what does the paxil (paroxetine) black box warning mean?

it is the antidepressant class warning with one line added. in pooled placebo-controlled trials of antidepressants, suicidal thoughts and behaviors increased in people under 25: 14 extra cases per 1,000 treated under 18 and 5 extra at 18 to 24, against 1 fewer at 25 to 64 and 6 fewer at 65 and over. the added line is that paroxetine is not approved for anyone under 18: three trials in 752 children did not show it worked for depression, and the label lists self-harm and suicidal thoughts among the reactions reported in those trials at twice the placebo rate. the warning says to watch everyone starting the drug, especially in the first months and after dose changes. it is not an instruction to stop.

is paroxetine safe in pregnancy?

the label carries a warning the other ssris do not: meta-analyses have linked first-trimester exposure to a less than twofold increase in heart defects, with two to two and a half times the rate of one specific defect (right ventricular outflow tract), and it says paroxetine should be started in women who are or intend to become pregnant only after other options are considered. the label also names the studies’ limitations, and a large medicaid study found the specific paroxetine heart signal fell to a relative risk of 1.07 once the comparison was restricted to women with depression. late in pregnancy the class risks apply: newborn adaptation problems and persistent pulmonary hypertension. on the other side, the label cites a study in which women who stopped antidepressants in pregnancy relapsed more often. the fda warned in 2005, 2006 and 2011; the history is on a linked page. this is a decision to make with a prescriber before, not after, a change.

why is paroxetine not approved for children or teenagers?

because three placebo-controlled trials in 752 paroxetine-treated children and adolescents with depression did not show it worked, and in those trials emotional lability, including self-harm and suicidal thoughts, hostility and agitation were reported at twice the placebo rate. the label states "paroxetine is not approved for use in pediatric patients" in the boxed warning itself. an independent 2015 re-analysis of one of those trials, study 329 in 275 adolescents, found paroxetine no more effective than placebo and associated with clinically significant increases in suicidal ideation and behaviour. fluoxetine’s label carries a depression approval from age 8, explained on its own page on this site; paroxetine’s does not [fluoxetine-explainer].

what are the sexual side effects of paroxetine?

the highest counted on this site. in the label’s pooled trials, among 1,446 men on paroxetine against 1,042 on placebo: ejaculatory disturbance 13% to 28% vs 0% to 2%, decreased libido 6% to 15% vs 0% to 5%, impotence 2% to 9% vs 0% to 3%. among 1,822 women against 1,340: decreased libido 0% to 9% vs 0% to 2%, orgasmic disturbance 2% to 9% vs 0% to 1%. the label says these figures probably understate the problem because people do not volunteer it. a questionnaire-based meta-analysis ranked paroxetine fourth of ten antidepressants for sexual dysfunction, behind sertraline, venlafaxine and citalopram. in 2019 the european regulator required a warning that sexual dysfunction has in some cases persisted after stopping; the us label does not carry it.

what happens if you stop paroxetine suddenly?

the label lists nausea, sweating, low mood, irritability, agitation, dizziness, electric-shock sensations, tremor, anxiety, confusion, headache, lethargy, emotional lability, insomnia, hypomania, tinnitus and seizures after stopping serotonergic antidepressants, particularly abruptly, and recommends gradual reduction whenever possible. its own trials tapered by 10 mg a week with a final week at 20 mg. how long withdrawal lasts varies: systematic reviews put the frequency anywhere from about 15% to 56% depending on method, and the uk’s nice guideline says it can take months and that the steps should get smaller as the dose gets lower. the schedule is a prescriber conversation.

can you take paroxetine with tamoxifen?

the label says to consider a different antidepressant. paroxetine irreversibly blocks cyp2d6, the enzyme that turns tamoxifen into its active form. the label reports that some studies found tamoxifen less effective at preventing relapse or death with paroxetine alongside, one suggesting the risk rises with longer overlap, and that other studies did not find the risk; its instruction is to "consider using an alternative antidepressant with little or no cyp2d6 inhibition" when tamoxifen is used. the ontario study behind the first finding estimated one extra breast-cancer death per 19.7 women with the median overlap. that is a question for the oncologist and the prescriber together.

what are the most common paroxetine side effects?

in the depression trials: nausea 26% vs 9% on placebo, somnolence 23% vs 9%, dry mouth 18% vs 12%, asthenia 15% vs 6%, constipation 14% vs 9%, dizziness 13% vs 6%, insomnia 13% vs 6%, sweating 11% vs 2%, tremor 8% vs 2%, plus the sexual side effects above. in the anxiety trials nausea ran 19% to 25%, somnolence 15% to 24% and abnormal ejaculation 13% to 28%. side effects rose with dose from 10 to 40 mg, and 20% of people in the depression trials stopped because of them, against 9% to 16% in the anxiety conditions. weight gain is listed as frequent.

Medication approval journey

paroxetine (Paxil)

Approved for Major depressive disorder in adults

page first published 2026-08-31 · last updated 2026-10-06

Also answers to “the subpoena magnet” — its codename in our videos and posts. The full codebook

FDA approvedSSRI antidepressantTaper risk: moderate

How this page reads

The page as written, with the clinical terms

Plain language uses the same facts, the same numbers and the same caveats. It does not leave anything out. Quotes from the FDA label are never reworded in either setting.

Before you stop or change your paroxetine dose

Stopping abruptly causes withdrawal effects that can be severe

Stopping or reducing an SSRI too fast commonly causes withdrawal symptoms — dizziness, electric-shock sensations, insomnia, agitation. These are often more severe and far longer-lasting than the "one to two weeks" many people are told. Any change should be a slow, prescriber-supervised taper.

what the paroxetine warnings mean

label facts below are read from the paroxetine tablets FDA label on DailyMed, set id 598ece4d-0ac4-993b-e063-6294a90a1d34, version 1, effective august 21, 2026 (zydus, repackaged by coupler; a generic label in the 2024 reformatted layout), the version the dossier above cites. the section numbers in brackets are the label’s own. every source is linked at the end of this section. last checked 2026-10-06.

paroxetine was approved on december 29, 1992 for depression, and later for obsessive-compulsive disorder, panic disorder, social anxiety disorder, generalized anxiety disorder and post-traumatic stress disorder, all in adults [label 1; approval]. about 1.7 million americans filled a prescription for it in 2024, down from about 2.4 million in 2014, a fall of about half that is larger than the survey’s margin of error [meps]. it is an ssri, and three things set its label apart from the other ssris on this site: it is the only one not approved for anyone under 18, and its boxed warning says so in its first lines; it is the only one with a pregnancy warning of its own, on heart defects after first-trimester exposure; and it is the strongest inhibitor among them of the liver enzyme cyp2d6, which is where its tamoxifen warning and two contraindicated antipsychotics come from. its label carries one boxed warning and thirteen numbered warnings. this section says what each means in plain words, what the trials behind each of the six uses measured and for how long, including the trials that failed, which side effects the label counts and how often, and what independent research has found since, in both directions. every number is from the label or from a study linked at the end.

the legal and safety record for this drug has its own page on this site and is not repeated here [legal-record]. this is information to take to a prescriber. nothing here is a reason to start, stop or change a dose on your own; the label itself says to reduce the dose gradually rather than stop abruptly whenever possible, and the medication guide says stopping suddenly "may cause you to have serious side effects".

the boxed warning: suicidal thoughts and behaviors, by age, and the pediatric line

the boxed warning says antidepressants increased the risk of suicidal thoughts and behaviors in pediatric and young adult patients in short-term studies, that everyone on one should be monitored closely for worsening, and, in its own words, that "paroxetine is not approved for use in pediatric patients" [label, boxed warning and 5.1]. the number behind the class warning is pooled placebo-controlled trials of antidepressants covering about 77,000 adults and 4,500 children. the label states the difference against placebo per 1,000 people treated: under 18, 14 additional cases; 18 to 24, 5 additional; 25 to 64, 1 fewer; 65 and older, 6 fewer [label 5.1, table 2]. it notes "considerable variation in risk" among drugs, that the risk beyond four months is unknown, and that prescribers should consider changing the regimen, including possibly stopping, in anyone whose depression is persistently worse [label 5.1].

the pediatric line has its own record in the label. effectiveness "was not demonstrated in three placebo-controlled trials in 752 paroxetine-treated pediatric patients" with depression [label 8.4]. in those trials the reactions reported in at least 2% of children and at twice the placebo rate were "emotional lability (including self-harm, suicidal thoughts, attempted suicide, crying, and mood fluctuations), hostility, decreased appetite, tremor, sweating, hyperkinesia, and agitation", and on stopping, even with a taper, emotional lability "including suicidal ideation, suicide attempt, mood changes, and tearfulness", nervousness, dizziness, nausea and abdominal pain [label 8.4]. one of those three trials was study 329; an independent re-analysis of its data in 275 adolescents, published in 2015, found paroxetine no more effective than placebo and associated with clinically significant increases in suicidal ideation and behaviour, reversing the original publication’s conclusions [le-noury-2015]. the age boundary above 24, which the pooled trials showed and the label prints, applies to paroxetine as to the other antidepressants [age-ceiling].

the pregnancy warning, in the label’s words, and who warned when

warning 5.4, embryofetal toxicity, is the one no other ssri label on this site carries. in the label’s words: "based on meta-analyses of epidemiological studies, exposure to paroxetine in the first trimester of pregnancy is associated with a less than 2-fold increase in the rate of cardiovascular malformations among infants. for women who intend to become pregnant or who are in their first trimester of pregnancy, paroxetine should be initiated only after consideration of the other available treatment options" [label 5.4]. the detail is in section 8.1: individual studies "have reported inconsistent findings"; meta-analyses of cohort studies published between 1996 and 2017 found a less than twofold increase in cardiovascular malformations overall, two of them an approximately 2 to 2.5-fold increase in right ventricular outflow tract defects, and one a pooled odds ratio of 2.38 (confidence interval 1.14 to 4.97) for atrial septal defect; the label names the limitations, "potential confounding by indication, depression severity and potential exposure misclassification" [label 8.1]. the background risk of major birth defects in the general population is 2% to 4% [label 8.1].

the rest of the pregnancy section is the ssri class text. exposure in the month before delivery is associated with a less than twofold increase in postpartum haemorrhage [label 5.5, 8.1]. babies exposed late in pregnancy have needed prolonged hospital care, breathing support and tube feeding, with symptoms the label says are consistent with either a toxic effect or a withdrawal syndrome, and there is a risk of persistent pulmonary hypertension of the newborn, which occurs in 1 to 2 per 1,000 births generally [label 8.1]. on the other side of the ledger the label puts untreated depression: in a study of 201 women, those who stopped antidepressants in pregnancy relapsed more often than those who continued, and the label says to "consider the risk of untreated depression when discontinuing or changing treatment" [label 8.1]. the label also says paroxetine may affect sperm quality and that it is not known whether that reverses [label 8.3].

the regulator history behind this warning is already on this site and is linked rather than repeated: the fda’s december 2005 advisory on paroxetine and heart defects, its july 2006 advisory on all ssris and newborn lung pressure, its december 2011 update calling the lung evidence conflicting and telling doctors not to change practice, and the uk regulator’s notices, each read on the agency’s own page [who-warned]. two library rows sit beside the label’s meta-analyses: a 2014 study of 949,504 medicaid pregnancies in which the paroxetine signal for right ventricular outflow tract defects fell to a relative risk of 1.07 (0.59 to 1.93) once the comparison was restricted to women with depression and fully adjusted [huybrechts-2014], and the 2014 federal rule that retired the pregnancy letter categories because the letters were being read as a ranking, which is why the current label carries prose rather than a letter [plr-2014]. the label and the medicaid study disagree on whether the heart signal survives adjustment, and this page does not settle it.

cyp2d6: tamoxifen, two contraindicated antipsychotics, and the other interactions

paroxetine irreversibly inhibits the liver enzyme cyp2d6, and the label draws three consequences. tamoxifen, the breast-cancer drug, is a prodrug that cyp2d6 turns into its active form; warning 5.11 says some studies have shown that tamoxifen’s efficacy, "as measured by the risk of breast cancer relapse/mortality, may be reduced" with paroxetine, that one study suggests the risk rises with longer co-administration, and that "other studies have failed to demonstrate such a risk"; it tells prescribers to consider an antidepressant with little or no cyp2d6 inhibition when tamoxifen is used [label 5.11, 7]. the study behind the first half followed 2,430 women in ontario on tamoxifen and an ssri and estimated one additional breast-cancer death within five years for every 19.7 women whose tamoxifen overlapped with paroxetine for the median 41% of the time, with no such signal for other ssris [kelly-2010]; the label’s own record of the dispute is on a library row [paxil-cr-label].

thioridazine and pimozide, two antipsychotics that prolong the qt interval, are contraindicated with paroxetine because it raises their blood levels [label 4, 5.3]. and a list of other cyp2d6 drugs may need a lower dose alongside paroxetine and a higher one if paroxetine stops: the label names propafenone, flecainide, atomoxetine, desipramine, dextromethorphan, metoprolol, nebivolol, perphenazine, tolterodine, venlafaxine and risperidone [label 7, table 9]. a case of severe low blood pressure when paroxetine was added to long-term metoprolol is in the post-marketing reports [label 6.2]. the hiv combination fosamprenavir/ritonavir lowers paroxetine levels significantly [label 7]. maois, including the antibiotic linezolid and intravenous methylene blue, are contraindicated within 14 days either way because of serotonin syndrome [label 4, 2.6].

the discontinuation syndrome, and the taper the trials used

warning 5.7 lists what has followed stopping serotonergic antidepressants, particularly abruptly: nausea, sweating, low mood, irritability, agitation, dizziness, sensory disturbances such as electric-shock sensations, tremor, anxiety, confusion, headache, lethargy, emotional lability, insomnia, hypomania, tinnitus and seizures, and it recommends a gradual reduction rather than abrupt cessation whenever possible [label 5.7, 2.7]. unusually, it then says what gradual meant in its own trials: "during clinical trials of gad and ptsd, gradual decreases in the daily dose by 10 mg/day at weekly intervals followed by 1 week at 20 mg/day was used before treatment was discontinued" [label 5.7]. in children, abnormal dreams, pins and needles and dizziness were reported on stopping, and the emotional lability above [label 5.7, 8.4]. "withdrawal syndrome" and "drug dependence" appear in the label’s list of rare reactions seen in the trials [label 6.1].

what the label does not say is how long withdrawal lasts or how often it happens, and the research section below has the two regulators and the two systematic reviews that address that. the label makes no comparison with other ssris on this point, and this page does not either.

the other warnings, in plain words

the warnings section runs from 5.1 to 5.13, with the pregnancy warning revised in november 2024 [label, recent major changes]. beyond those above:

serotonin syndrome [5.2]: too much serotonin signalling, with agitation or confusion, racing heart, sweating, fever, tremor, muscle jerking and overactive reflexes. the risk rises with triptans, tricyclics, fentanyl, lithium, tramadol, meperidine, methadone, tryptophan, buspirone, amphetamines and st john’s wort, and the combination with an maoi is contraindicated; it can also happen on paroxetine alone. a 2003 study of 2,222 overdoses defined the seven signs clinicians use [dunkley-2003].
bleeding [5.5]: drugs that block serotonin reuptake impair platelets. the label reports gastrointestinal bleeding in case-control and cohort studies, bruising, nosebleeds and life-threatening haemorrhage, and asks for the inr to be monitored carefully in anyone on warfarin. a meta-analysis of 15 case-control and 4 cohort studies put the odds of an upper gastrointestinal bleed at about 1.66 times on an ssri alone and 4.25 times with an nsaid as well, with one extra bleed per roughly 3,177 low-risk people a year and per 881 high-risk [anglin-2014].
mania [5.6]: hypomania or mania in about 1% of people with unipolar depression on paroxetine against 1.1% on active comparators and 0.3% on placebo; the label says to screen for a personal or family history of bipolar disorder first [2.4].
seizures [5.8]: 0.1% in the trials, with seizure disorders excluded; the label says to stop in anyone who develops one.
glaucoma [5.9]: pupil dilation can trigger an angle-closure attack in an eye with untreated narrow angles; cases have been reported with paroxetine, and the label says to avoid antidepressants in that situation.
low sodium [5.10]: hyponatremia, with cases below 110 mmol/l, usually from inappropriate antidiuretic hormone secretion. signs are headache, poor concentration, confusion, weakness and unsteadiness; severe cases have included seizures, coma, respiratory arrest and death. older people and people on diuretics are at greater risk.
bone fracture [5.12]: epidemiological studies have linked antidepressants including ssris to fractures; the label says there are multiple possible causes and "it is unknown to what extent fracture risk is directly attributable to ssri treatment". an english case-control study of 58,769 people with dementia found the highest cumulative exposure to strongly anticholinergic drugs, a list that includes paroxetine among the antidepressants, associated with about 49% higher odds of dementia, 29% for the antidepressant class; early dementia leading to prescriptions is the standard alternative explanation [coupland-2019].
sexual dysfunction [5.13]: delayed or failed ejaculation, low desire and erectile dysfunction in men, low desire and delayed or absent orgasm in women, with the instruction to ask before starting and during treatment because people do not volunteer it. the numbers, which are the highest on this site, are in the side-effects section.
post-marketing reports [6.2]: liver enzyme rises up to deaths from liver necrosis, stevens-johnson syndrome and toxic epidermal necrolysis, dress, movement disorders including akathisia and oculogyric crisis (the last with pimozide), status epilepticus, torsade de pointes, pulmonary hypertension, premature births, and raised prolactin with milk production.

the six approved uses, their trials, and the trials that failed

depression rests on six placebo-controlled studies in adults aged 18 to 73, which the safety section describes as 6-week trials of 20 to 50 mg, and a randomised-withdrawal study in which responders to 8 weeks of open-label paroxetine were kept on it or switched to placebo for up to a year, relapsing at 15% against 39% [label 14.1; 6.1]. the label gives no effect sizes or sample sizes for the depression trials, which is unusual among the labels on this site; the research section has what the published and unpublished trials showed when re-examined.

obsessive-compulsive disorder rests on two 12-week trials: a fixed-dose study in which 40 and 60 mg reduced the yale-brown scale by about 6 and 7 points against about 3 on placebo while 20 mg did not separate, and a flexible-dose study against clomipramine and placebo [label 14.2, table 10]. panic disorder rests on three 10- to 12-week trials: in the fixed-dose one only 40 mg beat placebo, with 76% panic-free against 44%; in the flexible-dose trials 51% against 32% were panic-free, and 33% against 14% reached zero or one attack when everyone also had cognitive behavioural therapy [label 14.3]. social anxiety rests on three 12-week trials, with 69% against 29% and 77% against 42% rated much or very much improved, and no additional benefit above 20 mg [label 14.4]. generalized anxiety rests on two positive 8-week trials; "a third study ... did not demonstrate statistically significant superiority of paroxetine over placebo" on its primary outcome, and a 24-week relapse-prevention trial in 566 responders found fewer relapses on the drug [label 14.5]. post-traumatic stress disorder rests on two positive 12-week trials, with a third positive on the clinician scale but not on the global-improvement responder measure; two thirds of the participants were women, and the mean duration of symptoms was 13 years [label 14.6]. for all six uses the maximum doses are 40 to 60 mg, and the label says 20 mg is enough for social and generalized anxiety [label 2.2, 2.3].

the arithmetic that matters: the efficacy trials were 6 to 12 weeks, the one controlled long-term measurement in depression was a randomised withdrawal of up to a year, children were studied in three trials and never approved, and about 1.7 million americans a year still take it, half as many as a decade ago.

side effects, with the label’s numbers

in the 6-week depression trials (421 on paroxetine against 421 on placebo): nausea 26% against 9%, somnolence 23% against 9%, dry mouth 18% against 12%, asthenia 15% against 6%, constipation 14% against 9%, dizziness 13% against 6%, insomnia 13% against 6%, ejaculatory disturbance 13% against 0% and other male genital disorders 10% against 0% (percentages corrected for sex), sweating 11% against 2%, tremor 8% against 2%, decreased appetite 6% against 2%, yawning 4% against 0%, decreased libido 3% against 0%, female genital disorders 2% against 0% [label 6.1, table 4]. in the obsessive-compulsive, panic and social anxiety trials: nausea 23% to 25% against 7% to 17%, somnolence 19% to 24% against 5% to 11%, insomnia 18% to 24% against 10% to 16%, tremor 9% to 11% against 1%, abnormal ejaculation 21% to 28% against 1%, decreased libido 7% to 12% against 1% to 4%, impotence 5% to 8% against 0% to 1% [label 6.1, table 5]. in the generalized anxiety and post-traumatic stress trials: nausea 19% to 20% against 5% to 8%, somnolence 15% to 16% against 5%, abnormal ejaculation 13% to 25% against 2% [label 6.1, table 6].

the sexual numbers pooled across all six conditions, in the label’s own table: among 1,446 men on paroxetine against 1,042 on placebo, decreased libido 6% to 15% against 0% to 5%, ejaculatory disturbance 13% to 28% against 0% to 2%, impotence 2% to 9% against 0% to 3%; among 1,822 women against 1,340, decreased libido 0% to 9% against 0% to 2%, orgasmic disturbance 2% to 9% against 0% to 1% [label 6.1, table 8]. the label adds that these estimates "may underestimate their actual incidence" because patients and clinicians are reluctant to discuss them, and reports several cases of priapism [label 6.1].

side effects rose with dose in the fixed-dose depression trial: nausea 15%, 27%, 35% and 36% at 10, 20, 30 and 40 mg against 14% on placebo; somnolence 13% to 22% against 8%; tremor 0% to 15% against 0%; abnormal ejaculation 6% to 13% against 0% [label 6.1, table 7]. dropouts for side effects were 20% of the 6,145 people in the depression trials, 16.1% in social anxiety, 11.8% in ocd, 9.4% in panic, 10.7% in generalized anxiety and 11.7% in ptsd, with nausea, somnolence, asthenia and abnormal ejaculation the leading drug-related reasons [label 6.1, table 3]. weight gain is listed as a frequent reaction [label 6.1]. in overdose, seizures that may be delayed, coma, qt prolongation and serotonin syndrome have been reported [label 10].

what the research says, in both directions

in favour. in the network meta-analysis of 522 trials covering 116,477 adults with depression, paroxetine beat placebo and was among the more effective antidepressants in head-to-head comparisons, though not among the best tolerated [cipriani-2018]. a cochrane review of 115 trials in 26,134 people found it neither clearly better nor clearly worse than other antidepressants overall [purgato-2014]. in an 8-week randomised trial of 120 adults with panic disorder, paroxetine and clonazepam both reduced attacks to a fraction of an attack a week [nardi-2011]. the label’s own relapse-prevention data, 15% against 39% over a year, are the strongest numbers on this page, with the standard caution that a randomised withdrawal cannot separate relapse from withdrawal effects [label 14.1].

against, or smaller than it sounds. when 40 published and unpublished placebo-controlled trials in 6,391 adults were re-examined together, paroxetine was not superior to placebo on a combined measure of effectiveness and acceptability, people on it stopped for side effects 77% more often, and suicidal tendencies were 2.55 times as likely [barbui-2008]. the cochrane review found some evidence it was less effective than citalopram on early response [purgato-2014]. in a 16-week randomised trial of 240 people comparing paroxetine with cognitive therapy, each prior course of antidepressants predicted a weaker response to paroxetine and no weaker response to therapy, a secondary analysis its authors asked to have replicated [leykin-2007]. in adolescents, the re-analysis of study 329 is above [le-noury-2015], and the network meta-analysis of 34 pediatric trials rated the evidence very low across the board [cipriani-2016]. in the panic trial, 95% of people on paroxetine reported side effects against 73% on clonazepam [nardi-2011].

the warnings, measured. sexual dysfunction: when it is asked about with a questionnaire rather than waited for, a meta-analysis found treatment-emergent sexual dysfunction in 25.8% to 80.3% of people across antidepressants, with paroxetine fourth of ten drugs significantly worse than placebo, behind sertraline, venlafaxine and citalopram [serretti-2009]; in 2019 the european regulator ordered every maker of paroxetine and nine other ssris and snris to warn that sexual dysfunction has in some cases continued after stopping, a warning the us label does not carry [prac-2019]. discontinuation: how often withdrawal happens across antidepressants is disputed between about 15% and 56% depending on method, and that it happens is not [davies-read-2019]; the uk’s nice guideline says withdrawal can take months, that a medicine should not be stopped abruptly except in exceptional circumstances, and that the taper should be a slow, stepwise reduction proportionate to the current dose, so the cuts get smaller near the end, which is a different shape from the fixed 10 mg weekly steps the label’s trials used [nice-ng215]; the royal college of psychiatrists in 2019 asked for tapering over potentially several months at a rate the patient can tolerate, while stating that antidepressant use is not dependence in the addiction sense [rcpsych-2019].

questions to ask your prescriber about paroxetine

each one comes from a warning or a study above. they are questions, not instructions; the answers depend on a history nobody here has.

i am pregnant, or might become pregnant. the label says paroxetine should be started only after other options are considered, and it also says stopping antidepressants in pregnancy raised relapse in a study. what are the options, and what did we weigh?
i take tamoxifen. the label says to consider an antidepressant with little or no cyp2d6 inhibition. was that considered?
the label counts ejaculatory disturbance in 13% to 28% of men and orgasm problems in 2% to 9% of women, and says those figures probably understate it. it also tells prescribers to ask. i would like to be asked.
the label’s trials tapered by 10 mg a week with a week at 20 mg, and nice says to make the steps smaller near the end. if we ever reduce or stop, what would the schedule be, and how do we tell withdrawal from relapse?
the label says 20 mg is enough for social anxiety and generalized anxiety and that side effects rise with dose. if i am above 20 mg for either, what is the higher dose for?
i take metoprolol, risperidone, atomoxetine, flecainide, propafenone, dextromethorphan, warfarin, an nsaid or aspirin. the label names each of these. is anything i take on that list?
if i am under 25: the boxed warning is about me, and the label says paroxetine is not approved under 18 at all. who is watching for worsening in the first months, and what exactly should they watch for?
if i am over 65: the label says to start at 10 mg and not exceed 40, and that older people are at greater risk of low sodium and of fractures. when was my sodium last checked?

questions people ask

what does the paxil (paroxetine) black box warning mean?: it is the antidepressant class warning with one line added. in pooled placebo-controlled trials of antidepressants, suicidal thoughts and behaviors increased in people under 25: 14 extra cases per 1,000 treated under 18 and 5 extra at 18 to 24, against 1 fewer at 25 to 64 and 6 fewer at 65 and over. the added line is that paroxetine is not approved for anyone under 18: three trials in 752 children did not show it worked for depression, and the label lists self-harm and suicidal thoughts among the reactions reported in those trials at twice the placebo rate. the warning says to watch everyone starting the drug, especially in the first months and after dose changes. it is not an instruction to stop.
is paroxetine safe in pregnancy?: the label carries a warning the other ssris do not: meta-analyses have linked first-trimester exposure to a less than twofold increase in heart defects, with two to two and a half times the rate of one specific defect (right ventricular outflow tract), and it says paroxetine should be started in women who are or intend to become pregnant only after other options are considered. the label also names the studies’ limitations, and a large medicaid study found the specific paroxetine heart signal fell to a relative risk of 1.07 once the comparison was restricted to women with depression. late in pregnancy the class risks apply: newborn adaptation problems and persistent pulmonary hypertension. on the other side, the label cites a study in which women who stopped antidepressants in pregnancy relapsed more often. the fda warned in 2005, 2006 and 2011; the history is on a linked page. this is a decision to make with a prescriber before, not after, a change.
why is paroxetine not approved for children or teenagers?: because three placebo-controlled trials in 752 paroxetine-treated children and adolescents with depression did not show it worked, and in those trials emotional lability, including self-harm and suicidal thoughts, hostility and agitation were reported at twice the placebo rate. the label states "paroxetine is not approved for use in pediatric patients" in the boxed warning itself. an independent 2015 re-analysis of one of those trials, study 329 in 275 adolescents, found paroxetine no more effective than placebo and associated with clinically significant increases in suicidal ideation and behaviour. fluoxetine’s label carries a depression approval from age 8, explained on its own page on this site; paroxetine’s does not [fluoxetine-explainer].
what are the sexual side effects of paroxetine?: the highest counted on this site. in the label’s pooled trials, among 1,446 men on paroxetine against 1,042 on placebo: ejaculatory disturbance 13% to 28% vs 0% to 2%, decreased libido 6% to 15% vs 0% to 5%, impotence 2% to 9% vs 0% to 3%. among 1,822 women against 1,340: decreased libido 0% to 9% vs 0% to 2%, orgasmic disturbance 2% to 9% vs 0% to 1%. the label says these figures probably understate the problem because people do not volunteer it. a questionnaire-based meta-analysis ranked paroxetine fourth of ten antidepressants for sexual dysfunction, behind sertraline, venlafaxine and citalopram. in 2019 the european regulator required a warning that sexual dysfunction has in some cases persisted after stopping; the us label does not carry it.
what happens if you stop paroxetine suddenly?: the label lists nausea, sweating, low mood, irritability, agitation, dizziness, electric-shock sensations, tremor, anxiety, confusion, headache, lethargy, emotional lability, insomnia, hypomania, tinnitus and seizures after stopping serotonergic antidepressants, particularly abruptly, and recommends gradual reduction whenever possible. its own trials tapered by 10 mg a week with a final week at 20 mg. how long withdrawal lasts varies: systematic reviews put the frequency anywhere from about 15% to 56% depending on method, and the uk’s nice guideline says it can take months and that the steps should get smaller as the dose gets lower. the schedule is a prescriber conversation.
can you take paroxetine with tamoxifen?: the label says to consider a different antidepressant. paroxetine irreversibly blocks cyp2d6, the enzyme that turns tamoxifen into its active form. the label reports that some studies found tamoxifen less effective at preventing relapse or death with paroxetine alongside, one suggesting the risk rises with longer overlap, and that other studies did not find the risk; its instruction is to "consider using an alternative antidepressant with little or no cyp2d6 inhibition" when tamoxifen is used. the ontario study behind the first finding estimated one extra breast-cancer death per 19.7 women with the median overlap. that is a question for the oncologist and the prescriber together.
what are the most common paroxetine side effects?: in the depression trials: nausea 26% vs 9% on placebo, somnolence 23% vs 9%, dry mouth 18% vs 12%, asthenia 15% vs 6%, constipation 14% vs 9%, dizziness 13% vs 6%, insomnia 13% vs 6%, sweating 11% vs 2%, tremor 8% vs 2%, plus the sexual side effects above. in the anxiety trials nausea ran 19% to 25%, somnolence 15% to 24% and abnormal ejaculation 13% to 28%. side effects rose with dose from 10 to 40 mg, and 20% of people in the depression trials stopped because of them, against 9% to 16% in the anxiety conditions. weight gain is listed as frequent.

sources for this section

the library rows link to our one-page summary of each study, with the original paper, its funding and our evidence grade on that page.

we went and looked

confirmeddoes the aripiprazole label call for half the dose with a strong cyp2d6 inhibitor like bupropion?said on PsychEd: Educational Psychiatry Podcast, 2026-09-01; the speaker has the label right. the current aripiprazole prescribing information tells prescribers to "administer half of usual dose" when aripiprazole is taken with a strong CYP2D6 inhibitor, and FDA's own reference table lists bupropion as a strong CYP2D6 inhibitor.

every check quotes one sentence, links the original at the timestamp, and takes each assertion to the primary record. the full list: /checked.

How long the paroxetine trials actually ran

The longest trial behind the paroxetine approval ran 52 weeks (about 12 months).

The gap between how long the trials ran and how long people actually take these medications is the single most important thing on this page. It is not evidence that longer use is unsafe or ineffective. It is evidence that longer use was not what got tested.

People often stay on these drugs for far longer than the trials ran. The trials behind the approval ran 6 weeks.

That does not mean longer use is unsafe. It does not mean the drug stops working either. It means longer use was never the thing that got tested.

If you have been on it a long time, that is a fair thing to ask your prescriber about.

Boxed warning: suicidal thoughts and behaviors

The strongest warning the FDA puts on a label, often called a black box warning. It is reproduced below word for word, not summarised.

This is the strongest warning the FDA can put on a drug label. Some people call it a black box warning. We copied it word for word. We did not shorten it and we did not soften it.

The warning below is about suicidal thoughts. In short studies, antidepressants raised the risk of suicidal thoughts and behavior. That was in children and young adults. The label tells prescribers to watch every patient closely for this. It also tells them to watch for the illness getting worse.

The FDA’s own words follow, copied exactly.

BOXED WARNING WARNING: SUICIDAL THOUGHTS AND BEHAVIORS Antidepressants increased the risk of suicidal thoughts and behaviors in pediatric and young adult patients in short-term studies. Closely monitor all antidepressant-treated patients for clinical worsening, and for emergence of suicidal thoughts and behaviors [see Warnings and Precautions ( 5.1 )]. Paroxetine is not approved for use in pediatric patients [see Use in Specific Populations ( 8.4 )]. WARNING: SUICIDAL THOUGHTS AND BEHAVIORS See full prescribing information for complete boxed warning. Increased risk of suicidal thoughts and behavior in pediatric and young adult patients taking antidepressants. Closely monitor all antidepressant-treated patients for clinical worsening and emergence of suicidal thoughts and behaviors. Paroxetine is not approved for use in pediatric patients. ( 5.1 , 8.4 )

FDA label effective August 21, 2026 — read the full label on DailyMed

What paroxetine is FDA approved for, and the trials behind it

Approved for major depressive disorder in adults in 1992, under application NDA 020031. Source: FDA Drugs@FDA.

The label does not state how many patients enrolled in the 6 efficacy studies. Its depression incidence table draws on 421 patients on paroxetine and 421 on placebo from 6-week trials, and 6,145 patients received paroxetine across the wider depression trial program.

Six placebo-controlled studies in patients aged 18 to 73 established efficacy for depression; the label's safety section describes its depression trials as 6-week studies of paroxetine 20 mg to 50 mg daily. In a randomized-withdrawal study, responders to 8 weeks of open-label paroxetine continued paroxetine or switched to placebo for up to 1 year, with relapse in 15% versus 39%.

One gap first. The label does not say how many people enrolled in the 6 trials.

Its depression rate table draws on 421 people on the drug and 421 on a placebo, from 6-week trials. Across the wider depression program, 6,145 people took the drug.

The FDA approved paroxetine in 1992. The approved use was major depression in adults. The file number is NDA 020031. The source is Drugs@FDA, the FDA's own database.

Six trials against a placebo backed the depression approval. Patients were aged 18 to 73.

The label’s safety section describes those depression trials as 6-week studies. Doses ran from 20 mg to 50 mg a day.

One longer study took people who had done well over 8 open weeks. Some stayed on the drug and some moved to a placebo by chance, for up to 1 year. 15 out of 100 got ill again on the drug. 39 out of 100 got ill again on the placebo.

Trials: 6 placebo-controlled studies
Duration: 6 weeks
Compared against: Placebo

The efficacy of paroxetine as a treatment for major depressive disorder (MDD) has been established in 6 placebo-controlled studies of patients with MDD (aged 18 to 73).

Verbatim from the paroxetine FDA label, Clinical Studies section, effective August 21, 2026 — read the full label on DailyMed.

Side effects of paroxetine in the approval trials

Incidence among 421 patients on paroxetine versus 421 on placebo in 6-week placebo-controlled depression trials at 20 mg to 50 mg daily. The label cautions that trial rates cannot be directly compared with rates for other drugs and may not reflect the rates observed in practice.

Read the limits first. These numbers come from 6-week depression trials against a placebo. Doses ran from 20 mg to 50 mg a day. 421 people took the drug and 421 took a placebo.

The label warns that you cannot line these rates up against another drug’s rates. It warns that real life may not match them either.

Nausea (26% vs 9% on placebo)
Somnolence (23% vs 9%)
Dry mouth (18% vs 12%)
Asthenia (15% vs 6%)
Constipation (14% vs 9%)
Dizziness (13% vs 6%)
Insomnia (13% vs 6%)
Ejaculatory disturbance (13% vs 0%, male patients)
Diarrhea (12% vs 8%)
Sweating (11% vs 2%)
Tremor (8% vs 2%)
Decreased appetite (6% vs 2%)

Most common adverse reactions (≥ 5% and at least twice placebo) are abnormal ejaculation, asthenia, constipation, decreased appetite, diarrhea, dizziness, dry mouth, female genital disorder, impotence, infection, insomnia, libido decreased, male genital disorder, nausea, nervousness, somnolence, sweating, tremor, yawn.

From the paroxetine FDA label, Adverse Reactions section, effective August 21, 2026 — DailyMed.

More from the paroxetine label: warnings and precautions

Discontinuation syndrome
The label lists adverse reactions after stopping serotonergic antidepressants, particularly abruptly, including nausea, dizziness, sensory disturbances such as electric shock sensations, anxiety, insomnia, and seizures, and recommends gradual dose reduction rather than abrupt cessation. In GAD and PTSD trials the dose was stepped down by 10 mg per day at weekly intervals before stopping.
FDA label, Warnings and Precautions section.
Discontinuation for adverse reactions in trials
In depression clinical trials, 1,199 of 6,145 patients treated with paroxetine stopped treatment because of an adverse reaction. The reactions most often leading to dropout at roughly twice the placebo rate included nausea, somnolence, asthenia, tremor, and abnormal ejaculation.
FDA label, Adverse Reactions section.

How many Americans take paroxetine

Survey-based federal estimates, published with a lag of about two years. They count prescriptions filled, not outcomes.

These are federal survey numbers. They come out about two years late. They count prescriptions filled. They do not tell you if the drug helped.

7,173,802
prescriptions in the United States (2024), refills included: 1,681,958
people filling them (2024), ± 403,092 (95%)

Prescriptions are down 52% from 2014 to 2024, a change larger than the survey’s margin of error. Whatever you decide about paroxetine, you are deciding alongside about 1,681,958 other people in 2024.

Prescriptions by year: 2014 15,020,666 · 2015 14,808,496 · 2016 12,222,862 · 2017 11,670,683 · 2018 11,228,908 · 2019 9,783,756 · 2020 9,029,668 · 2021 7,482,340 · 2022 7,165,530 · 2023 9,577,611 · 2024 7,173,802.

Prescriptions are down 52 percent since 2014. About 1.7 million people in the United States took it in 2024.

Correction (29 September 2026). An earlier version of this section took its numbers from ClinCalc DrugStats, a third-party site that republishes this survey; we now compute them from AHRQ’s public-use files ourselves.

Source: AHRQ Medical Expenditure Panel Survey (MEPS), Prescribed Medicines public-use files, 2014 (HC-168A) through 2024 (HC-254A), computed by veisund. People with at least one purchase where the Multum generic name (RXDRGNAM) equals the drug, weighted by the year’s person weight; purchases include refills. Standard errors by Taylor linearization on VARSTR/VARPSU. AHRQ’s reliability rule: a year is not shown when fewer than 60 survey respondents reported the drug or the relative standard error is over 50%. AHRQ cautions that its 2018 questionnaire redesign and the 2020–2022 pandemic years make year-to-year comparisons less certain.

What people report to the FDA about paroxetine

Read this before the numbers.

Anyone can file an adverse event report — patients, doctors, manufacturers — and nobody verifies that the drug caused what was reported. There is no denominator: tens of millions of prescriptions generate reports at an unknowable rate, and reporting rises with news coverage, not necessarily with harm. Counts below are report volumes, not rates, and cannot be compared between drugs. The FDA itself says not to use this data to make medical decisions — we show it because you deserve to see what is in the public record, with its limits stated plainly.

Read this before the numbers.

Anyone can send a report to the FDA. Patients can. Doctors can. Drug makers can. Nobody checks that the drug caused what the report describes.

There is also no bottom number. Tens of millions of prescriptions turn into reports at a rate nobody knows. Reports go up when a drug is in the news, and that is not the same as more harm.

So these are counts of reports. They are not rates. You cannot use them to compare one drug with another. The FDA itself says not to use this data to make medical choices.

We show it because it is on the public record, and you should get to see it with its limits named.

91,207
reports mentioning paroxetine, all time: 62,242
filed as serious (a report-level flag covering every drug and outcome in the report) filed as serious. That flag covers the whole report, and one report can name more than one drug and more than one outcome.

Most-reported reactions

Drug ineffective6,311
Drug withdrawal syndrome6,210
Nausea6,139
Dizziness5,469
Fatigue5,430
Anxiety5,396
Headache4,832
Depression4,322
Diarrhoea3,923
Insomnia3,890

“Drug ineffective” ranking this high is worth noticing: a report of not being helped counts as an adverse event too, and people file them in large numbers.

It is worth noticing how high “drug ineffective” ranks. A report that the drug did not help counts as an adverse event too. People file a lot of them.

Source: FDA Adverse Event Reporting System (FAERS), via openFDA, openFDA dataset refreshed 2026-07-30 (the refresh date of the query endpoint, not the date reports run through).

Drug interactions listed on the paroxetine label

The FDA label’s interactions section, verbatim. A pharmacist checking your actual medication list beats any published list — including this one.

Below is the part of the FDA label that lists other drugs. It is copied word for word.

No list on the web can see what else you take. A pharmacist can. Ask one to check your full list.

Read the label’s interactions section

7 DRUG INTERACTIONS Table 9 presents clinically significant drug interactions with paroxetine. Table 9 Clinically Significant Drug Interactions with Paroxetine Monoamine Oxidase Inhibitors (MAOIs) Clinical Impact The concomitant use of SSRIs, including paroxetine, and MAOIs increases the risk of serotonin syndrome. Intervention Paroxetine is contraindicated in patients taking MAOIs, including MAOIs such as linezolid or intravenous methylene blue [see Dosage and Administration ( 2.5 ), Contraindications ( 4 ), Warnings and Precautions ( 5.2 )] . Examples selegiline, tranylcypromine, isocarboxazid, phenelzine, linezolid, methylene blue Pimozide and Thioridazine Clinical Impact Increased plasma concentrations of pimozide and thioridazine, drugs with a narrow therapeutic index, may increase the risk of QTc prolongation and ventricular arrhythmias. Intervention Paroxetine is contraindicated in patients taking pimozide or thioridazine [see Contraindications ( 4 )] . Other Serotonergic Drugs Clinical Impact The concomitant use of serotonergic drugs with paroxetine increases the risk of serotonin syndrome. Intervention Monitor patients for signs and symptoms of serotonin syndrome, particularly during treatment initiation and dosage increases. If serotonin syndrome occurs, consider discontinuation of paroxetine and/or concomitant serotonergic drugs [see Warnings and Precautions ( 5.2 )] . Examples other SSRIs, SNRIs, triptans, tricyclic antidepressants, opioids, lithium, tryptophan, buspirone, amphetamines and St. John's Wort. Drugs that Interfere with Hemostasis (antiplatelet agents and anticoagulants) Clinical Impact The concurrent use of an antiplatelet agent or anticoagulant with paroxetine may potentiate the risk of bleeding. Intervention Inform patients of the increased risk of bleeding associated with the concomitant use of paroxetine and antiplatelet agents and anticoagulants. For patients taking warfarin, carefully monitor the international normalized ratio [see Warnings and Precautions ( 5.5 )] . Examples aspirin, clopidogrel, heparin, warfarin Drugs Highly Bound to Plasma Protein Clinical Impact Paroxetine is highly bound to plasma protein. The concomitant use of paroxetine with another drug that is highly bound to plasma protein may increase free concentrations of paroxetine or other tightly-bound drugs in plasma. Intervention Monitor for adverse reactions and reduce dosage of paroxetine or other protein-bound drugs as warranted. Examples warfarin Drugs Metabolized by CYP2D6 Clinical Impact Paroxetine is a CYP2D6 inhibitor [see Clinical Pharmacology ( 12.3 )] . The concomitant use of paroxetine with a CYP2D6 substrate may increase the exposure of the CYP2D6 substrate. Intervention Decrease the dosage of a CYP2D6 substrate if needed with concomitant paroxetine use. Conversely, an increase in dosage of a CYP2D6 substrate may be needed if paroxetine is discontinued. Examples propafenone, flecainide, atomoxetine, desipramine, dextromethorphan, metoprolol, nebivolol, perphenazine, tolterodine, venlafaxine, risperidone. Tamoxifen Clinical Impact Concomitant use of tamoxifen with paroxetine may lead to reduced plasma concentrations of the active metabolite (endoxifen) and reduced efficacy of tamoxifen Intervention Consider use of an alternative antidepressant with little or no CYP2D6 inhibition [see Warnings and Precautions ( 5.11 )] . Fosamprenavir/Ritonavir Clinical Impact Co-administration of fosamprenavir/ritonavir with paroxetine significantly decreased plasma levels of paroxetine. Intervention Any dose adjustment should be guided by clinical effect (tolerability and efficacy). Drugs Highly Bound to Plasma Protein: Monitor for adverse reactions and reduce dosage of paroxetine or other protein-bound drugs (e.g., warfarin) as warranted. ( 7 ) Drugs Metabolized by CYP2D6: Reduce dosage of drugs metabolized by CYP2D6 as warranted. ( 7 ) Concomitant use with tamoxifen: Consider use of an alternative antidepressant with little or no CYP2D6 inhibition. ( 5.11 , 7 )

FDA label for paroxetine, effective August 21, 2026 — DailyMed.

Label history: when this label changes, it is recorded in What changed.

Who pays for paroxetine

Two claims datasets and one survey, covering different populations with different instruments — they cannot be reconciled by arithmetic, and where their sum crowds the all-payer estimate, that is a finding about the estimates rather than a percentage.

Medicare Part D · claims · 2024Medicaid · claims floor · 2024All-payer · survey · 2024Commercially insured adults · not publishedChildren · no per-drug data

Medicare Part D: Read the under-65 group correctly before the numbers: Medicare before 65 means the disabled and dual-eligible population — among the sickest, highest-need people in the program — and reading their utilization as typical adult use would be a category error. The beneficiary total sums brand-level rows, so treat it as an upper bound on distinct people. 721,377 beneficiaries filled 3,478,176 claims in 2024 — 594,298 aged 65 and over, and 127,079 under 65.
Medicaid: At least 1,219,170 prescriptions in 2024 — a floor, because 184 of 940 national data rows are suppressed for privacy and contribute zero. Medicaid covers more children than any insurer in the country and publishes no age split — how much of this number is pediatric use is not knowable from public data.
All payers (survey estimate): A survey estimate from 102 respondents, which we computed from AHRQ’s MEPS 2024 Prescribed Medicines file (people with at least one purchase, weighted; purchases include refills), puts the whole country at 7,173,802 purchases and 1,681,958 people in 2024. Subtracting the public programs from it would produce a number for everyone else — and we do not print that number, because subtracting a survey from claims counts manufactures precision that does not exist.
The population nobody counts: The commercially insured working-age adult — statistically, the likeliest person to be reading this page — is the one population with no public per-drug count anywhere. Private claims data exists and is sold, but nothing a patient can check is published. For children the record is thinner still: no public source counts pediatric use of paroxetine specifically; the closest the public record comes is condition-level treatment rates for children, which we have traced for one condition in how childhood ADHD got counted.

Sources: Medicare Part D Prescribers — by Geography and Drug, data.cms.gov, National rows, data year 2024 (published with ~17-month lag). Medicaid State Drug Utilization Data 2024, data.medicaid.gov, national aggregate rows. Retrieved 2026-09-01.

Who has been paying for paroxetine, 2019–2024

Medicare Part D

Claims, all ages — the one payer with a public per-drug series.

20192024

4,049,672 → 3,478,176 claims

Medicare under 65

Beneficiaries, not ordinary working-age adults: Medicare before 65 means the disabled and dual-eligible population. A gap is a suppressed split, not zero.

20192024

185,839 → 127,079 beneficiaries

Medicaid

A floor each year — small-cell rows are suppressed federally and excluded from the sum.

20192024

1,857,000 → 1,219,170 prescriptions (floor)

Commercially insured adults

No public dataset publishes per-drug commercial claims for any year — the absence is the finding, and it is rendered, not interpolated. The population this library most wants to inform is the one with no public per-drug series, in any year — insurers and PBMs hold that data privately. What the federal surveys do measure — how many adults take these medications at all — is on the adult-usage page.

Sources: Medicare Part D Prescribers by Geography and Drug (data.cms.gov, national rows, per data year); Medicaid State Drug Utilization Data (data.medicaid.gov, national rows). Claims, fills and beneficiaries are different units — compare each series only with itself.

The paroxetine approval, step by step

Step 1
What the approval was actually based on
Which studies did the FDA rely on, how long did they run, and who was in them?
The efficacy of PAXIL as a treatment for major depressive disorder (MDD) has been established in 6 placebo-controlled studies of patients with MDD (aged 18 to 73)... Patients who responded to PAXIL (HDRS total score <8) during an initial 8-week open-label treatment phase were then randomized to continue PAXIL or placebo, for up to 1 year. Patients treated with PAXIL demonstrated a statistically significant lower relapse rate during the withdrawal phase (15%) compared to those on placebo (39%).
FDA-approved labelling, 14 CLINICAL STUDIES 14.1 Major Depressive Disorder — read the label on DailyMed
Our reading
Six short trials plus a full year of randomised withdrawal is one of the better duration records here, and the relapse split — 15% against 39% — is a large effect. Read the design before you read the number. Randomised withdrawal takes people already doing well on the drug and switches some to placebo, so it cannot separate relapse of the illness from withdrawal from the medication. Paroxetine has the shortest half-life of the common SSRIs and the most severe documented discontinuation syndrome, which makes that ambiguity larger for this drug than for any other on this list.
Step 2
The approval
When was it approved, under what application, and by whose review?
Approved
December 29, 1992
Application
NDA020031
Review
STANDARD
Original sponsor
SmithKline Beecham (now GlaxoSmithKline)
Holds it now
Apotex
Label submissions since
64
Source: openFDA Drugs@FDA, original application ORIG-1
Step 3
What was added after it was on the market
Which warnings arrived only after millions of people were already taking it?
- PRAC recommendations on signals adopted at the 13-16 May 2019 PRAC meeting, section 1.3: SNRIs and SSRIs - persistent…
  27 years after approval
  On 16 May 2019 the EU drug regulator's safety committee ordered every manufacturer of citalopram, escitalopram, fluvoxamine, fluoxetine, paroxetine, sertraline, duloxetine, venlafaxine, desvenlafaxine, and milnacipran to add this warning within two months: "There have been reports of long-lasting sexual dysfunction where the symptoms have continued despite discontinuation of SSRIs/SNRI." The patient leaflet wording is "In some cases, these symptoms have continued after stopping treatment." This is a regulator acting on case reports and pharmacovigilance data, not a prevalence study — how often it happens is still unknown, and that uncertainty cuts both ways.
  Worth asking
  European regulators require a label warning that sexual side effects can persist after stopping an SSRI or SNRI — how would you and I tell that apart from the depression itself if it happened to me?
  PRAC recommendations on signals adopted at the 13-16 May 2019 PRAC meeting, section 1.3: SNRIs and SSRIs - persistent sexual dysfunction after drug withdrawal (EMA/PRAC/265212/2019). — Pharmacovigilance Risk Assessment Committee (PRAC), European Medicines Agency. Rapporteur: Gross-Martirosyan L. (2019)
- The paroxetine label carries the tamoxifen warning, the counter-evidence, and a contraindication with two antipsychotics
  33 years after approval
  A label is where the regulator and the manufacturer record what they accept about an interaction. On tamoxifen, this one records the dispute.
  Significant findings
  PAXIL CR prescribing information (Apotex, September 2025), section 5.11, Reduction of Efficacy of Tamoxifen: "Some studies have shown that the efficacy of tamoxifen, as measured by the risk of breast cancer relapse/mortality, may be reduced with concomitant use of paroxetine as a result of paroxetine's irreversible inhibition of CYP2D6 and lower blood levels of tamoxifen. One study suggests that the risk may increase with longer duration of coadministration. However, other studies have failed to demonstrate such a risk. When tamoxifen is used for the treatment or prevention of breast cancer, prescribers should consider using an alternative antidepressant with little or no CYP2D6 inhibition."
  Section 5.3, Drug Interactions Leading to QT Prolongation: "The CYP2D6 inhibitory properties of paroxetine can elevate plasma levels of thioridazine and pimozide. Since thioridazine and pimozide given alone produce prolongation of the QTc interval and increase the risk of serious ventricular arrhythmias, the use of PAXIL CR is contraindicated in combination with thioridazine and pimozide".
  The patient counselling section lists the serotonin syndrome partners in the same words as the other SSRI labels: "triptans, tricyclic antidepressants, opioids, lithium, tryptophan, buspirone, amphetamines, St. John's Wort, and with drugs that impair metabolism of serotonin (in particular, MAOIs, both those intended to treat psychiatric disorders and also others, such as linezolid)."
  Worth asking
  Paroxetine's distinguishing interaction is CYP2D6: it blocks the enzyme irreversibly, which is why the tamoxifen and antipsychotic warnings are on this label and not, in the same form, on sertraline's or citalopram's. Other CYP2D6-dependent drugs include some beta blockers, codeine and tramadol. If you take paroxetine, the question for a pharmacist is which of your other drugs are metabolised by CYP2D6.
  PAXIL CR (paroxetine hydrochloride) extended-release tablets: FDA-approved prescribing information, Warnings and Precautions 5.3 and 5.11, Drug Interactions 7.1 — Apotex Corp; US Food and Drug Administration (2025)
- The antidepressant suicidality warning has an age ceiling most people never hear about
  Antidepressants carry a boxed warning about suicidal thoughts and behaviour in children, adolescents and young adults. It came from pooling 24 short-term placebo-controlled trials of nine antidepressants in more than 4,400 young patients: suicidality was reported in about 4% on drug against 2% on placebo. There were no completed suicides in those trials.
  The part that usually gets lost is the age boundary, which is in the label itself. The studies did not show an increased risk above age 24, and in patients 65 and older the risk went in the other direction — it was reduced. The warning is real and it is specific, not a blanket statement about everyone who takes one.
  Worth asking
  Given my age, which side of that line am I on, what specifically should I or the people around me watch for in the first two months, and who do I call if it happens.
  PROZAC (fluoxetine) prescribing information — Boxed Warning: Suicidal Thoughts and Behaviors — U.S. Food and Drug Administration (approved labelling)
Step 4
What independent research has found since
What has been learned by people who were not selling it?
- Paroxetine taken alongside tamoxifen was linked to more breast cancer deaths; other SSRIs were not
  18 years after approval
  Tamoxifen is a prodrug: the liver enzyme CYP2D6 turns it into the compound that fights the cancer. Paroxetine is the strongest CYP2D6 blocker among the SSRIs. This study asked whether that chemistry shows up in survival.
  Significant findings
  Kelly and colleagues (BMJ, 2010) followed women in Ontario aged 66 or older treated with tamoxifen between 1993 and 2005 who had overlapping treatment with a single SSRI. "Of 2430 women treated with tamoxifen and a single SSRI, 374 (15.4%) died of breast cancer during follow-up (mean follow-up 2.38 years, SD 2.59)."
  "After adjustment for age, duration of tamoxifen treatment, and other potential confounders, absolute increases of 25%, 50%, and 75% in the proportion of time on tamoxifen with overlapping use of paroxetine (an irreversible inhibitor of CYP2D6) were associated with 24%, 54%, and 91% increases in the risk of death from breast cancer, respectively (P<0.05 for each comparison). By contrast, no such risk was seen with other antidepressants."
  The absolute estimate: "We estimate that use of paroxetine for 41% of tamoxifen treatment (the median overlap in our sample) would result in one additional breast cancer death within five years of cessation of tamoxifen for every 19.7 (95% confidence interval 12.5 to 46.3) patients so treated".
  Worth asking
  The paroxetine label now says, in section 5.11, that "other studies have failed to demonstrate such a risk" and tells prescribers to "consider using an alternative antidepressant with little or no CYP2D6 inhibition" when tamoxifen is used; that row is beside this one. The disagreement is real. What is not disputed is the mechanism and that alternatives exist. If you take tamoxifen and any antidepressant, the question is which enzyme it inhibits, and it is a question for the oncologist and the prescriber together, not for you to act on alone.
  Selective serotonin reuptake inhibitors and breast cancer mortality in women receiving tamoxifen: a population based cohort study — Kelly CM, Juurlink DN, Gomes T, Duong-Hua M, Pritchard KI, Austin PC, Paszat LF (2010)
- Cochrane 2021: newer antidepressants cut depression in children and teens by a small and unimportant amount
  29 years after approval
  The Cochrane review of newer antidepressants for depression in children and teenagers, and the one that disagreed with the Lancet reviews on fluoxetine.
  Significant findings
  Hetrick and colleagues pooled 26 randomised trials of newer antidepressants in 6 to 18 year olds with major depression, searched to March 2020. Older drugs (tricyclics and MAO inhibitors) were not included. On the clinician-rated CDRS-R scale (17 to 113 points), most drugs gave a "small and unimportant" drop in symptoms compared with placebo. Fluoxetine: 2.84 points (95% CI 1.56 to 4.12), moderate certainty. Sertraline (3.51), escitalopram (2.62), duloxetine (2.70) and paroxetine (1.43) were in the same range. Differences between most of the drugs were also "small and unimportant".
  No trial reported the two primary outcomes, diagnosis by clinical interview and suicide. Suicide-related outcomes were uncommon and the estimates wide. For fluoxetine the odds ratio was 1.27 (0.87 to 1.86): low certainty that it may at least slightly raise those odds. Most trials excluded young people at risk of suicide, so the review cannot speak for them.
  The other direction
  The review does not say these drugs are useless. Averages hide people who respond more. The authors say sertraline, escitalopram, duloxetine and fluoxetine "could be considered as a first option", with close monitoring of treatment effects and suicide-related outcomes, and that therapy, particularly CBT, remains important. They also say methodological shortcomings in the trials make the findings hard to interpret.
  Why it matters now
  Because it left out tricyclics, this review did not include a small 2006 trial of fluoxetine against nortriptyline. A 2026 re-analysis (Lyus and colleagues, in this library) found that trial is what made fluoxetine look larger in the two Lancet reviews, including Cipriani 2016 (also in this library). Without it, all three land close to this review's estimate.
  What this does not show
  It does not tell any family whether to start, keep or stop a medicine. If your child takes fluoxetine, this is a question for their prescriber, not a reason to stop.
  Worth asking
  What size of change should we expect, how will we measure it, and how will suicide-related thinking be watched?
  New generation antidepressants for depression in children and adolescents: a network meta-analysis — Hetrick SE, McKenzie JE, Bailey AP, Sharma V, Moller CI, Badcock PB, Cox GR, Merry SN, Meader N (2021)
- Prior antidepressant courses predicted a weaker response to paroxetine, and no weaker response to cognitive therapy
  15 years after approval
  The trial that asked whether a history of antidepressants blunts the next treatment, and found it depended on what the next treatment was.
  Significant findings
  Leykin, Amsterdam, DeRubeis and colleagues used a randomised trial of 240 patients with moderate-to-severe major depressive disorder, comparing paroxetine with cognitive therapy over 16 weeks. Each person's history of prior antidepressant exposure was assessed "with structured interviews, self-report, and medical records." The analysis used hierarchical linear models on the intent-to-treat sample.
  "After controlling for various demographic and clinical factors, more prior AD exposures predicted poor response to paroxetine therapy but not to CT, as measured by the Hamilton Rating Scale for Depression." Spelled out: "Whereas CT outcome was not significantly related to the number of prior AD exposures, a higher number of prior AD exposures was significantly associated with a lower response to paroxetine."
  The implication the authors draw is conditional: "If these findings are replicated in methodologically rigorous studies of paroxetine and other antidepressants, CT should be recommended, in preference to AD, for patients with multiple prior AD exposures."
  The other direction
  The word "if" is doing real work. The authors ask for replication in rigorous studies of paroxetine and other drugs before the finding should change what anyone is offered, and the abstract does not say that replication has happened. It is one drug, paroxetine, over 16 weeks. The trial was designed to compare two treatments, not to test the prior-exposure question, so this is a secondary analysis. And a history of many antidepressant courses may mark a depression that is harder to treat by any means; the therapy arm's flat result argues against that, but does not settle it.
  What this does not show
  It does not say cognitive therapy is better than paroxetine for everyone, or that anyone with a long medication history should stop a drug. It says the two treatments responded differently to one variable, in one trial. Which treatment fits a given person is a decision with a prescriber and, where therapy is on the table, a therapist.
  Worth asking
  Given how many antidepressants I have tried, is therapy on the table, whether or not we change the medication?
  Progressive resistance to a selective serotonin reuptake inhibitor but not to cognitive therapy in the treatment of major depression — Leykin Y, Amsterdam JD, DeRubeis RJ, Gallop R, Shelton RC, Hollon SD (2007)
- SSRIs raise the odds of a stomach bleed by about two thirds; with an NSAID as well, about fourfold
  22 years after approval
  Serotonin is needed for platelets to clump. Antidepressants that block its reuptake reduce the platelet store, and this meta-analysis measured what that does to the risk of bleeding from the stomach, alone and with an anti-inflammatory painkiller.
  Significant findings
  Anglin and colleagues (American Journal of Gastroenterology, 2014) searched to April 2013 and included "Fifteen case-control studies (including 393,268 participants) and four cohort studies". "There was an increased risk of upper GI bleeding with SSRI medications in the case-control studies (OR=1.66, 95% CI=1.44,1.92) and cohort studies (OR=1.68, 95% CI=1.13,2.50)."
  The absolute numbers: "The number needed to harm for upper GI bleeding with SSRI treatment in a low-risk population was 3,177, and in a high-risk population it was 881." And the combination: "The risk of upper GI bleeding was further increased with the use of both SSRI and NSAID medications (OR=4.25, 95% CI=2.82,6.42)."
  The authors' conclusion: "SSRI medications are associated with a modest increase in the risk of upper GI bleeding, which is lower than has previously been estimated. This risk is significantly elevated when SSRI medications are used in combination with NSAIDs, and physicians prescribing these medications together should exercise caution and discuss this risk with patients."
  Worth asking
  The number needed to harm is the one to hold on to: on an SSRI alone, one extra bleed for roughly every three thousand low-risk people, or every nine hundred high-risk. The combination with regular NSAID use is where the risk stops being modest. If you take an SSRI and buy ibuprofen or naproxen for pain, the question for the pharmacist is what the alternative is, and whether you count as high risk (prior ulcer, older age, anticoagulants, steroids).
  Risk of upper gastrointestinal bleeding with selective serotonin reuptake inhibitors with or without concurrent nonsteroidal anti-inflammatory use: a systematic review and meta-analysis — Anglin R, Yuan Y, Moayyedi P, Tse F, Armstrong D, Leontiadis GI (2014)
- Years of stacked anticholinergic drugs, including some antidepressants and antipsychotics, were linked to dementia
  27 years after approval
  Some interactions are not between two named drugs but between a property several of them share. Anticholinergic burden is the clearest example in psychiatry: tricyclic antidepressants, some antipsychotics, bladder drugs and older antihistamines all block the same receptor, and the effects add.
  Significant findings
  Coupland and colleagues (JAMA Internal Medicine, 2019) ran a nested case-control study in English general practice records: "58 769 patients with a diagnosis of dementia and 225 574 controls 55 years or older matched by age, sex, general practice, and calendar time." Exposure was the cumulative prescribed dose of 56 strongly anticholinergic drugs in the 1 to 11 years before diagnosis.
  "The adjusted OR for dementia increased from 1.06 (95% CI, 1.03-1.09) in the lowest overall anticholinergic exposure category (total exposure of 1-90 TSDDs) to 1.49 (95% CI, 1.44-1.54) in the highest category (>1095 TSDDs), compared with no anticholinergic drug prescriptions in the 1 to 11 years before the index date."
  By class, at the highest exposure: "anticholinergic antidepressants (adjusted OR [AOR], 1.29; 95% CI, 1.24-1.34), antiparkinson drugs (AOR, 1.52; 95% CI, 1.16-2.00), antipsychotics (AOR, 1.70; 95% CI, 1.53-1.90), bladder antimuscarinic drugs (AOR, 1.65; 95% CI, 1.56-1.75), and antiepileptic drugs (AOR, 1.39; 95% CI, 1.22-1.57)". "Associations were stronger in cases diagnosed before the age of 80 years. The population-attributable fraction associated with total anticholinergic drug exposure during the 1 to 11 years before diagnosis was 10.3%."
  Worth asking
  A case-control study cannot rule out that early, undiagnosed dementia leads to these prescriptions rather than the reverse, though the authors tested windows starting 3 and 5 years before diagnosis and the association held. What the study supports is a question, not a decision: how many of your medications are strongly anticholinergic, and is the total something anyone has added up? A pharmacist can score it. The clozapine label in this wave carries the short-term version of the same warning, anticholinergic toxicity when such drugs are combined.
  Anticholinergic Drug Exposure and the Risk of Dementia: A Nested Case-Control Study — Coupland CAC, Hill T, Dening T, Morriss R, Moore M, Hippisley-Cox J (2019)
- Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major…
  26 years after approval
  Across 522 randomised trials with 116,477 adults, paroxetine beat placebo and was among the more effective antidepressants in head-to-head comparisons, but it was not among the best tolerated.
  Worth asking
  Paroxetine scored well on effect but not on tolerability in this comparison - which of its side effects should I watch for most in the first month?
  Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder: a systematic review and network meta-analysis — Cipriani A, et al. (2018)
- Paroxetine versus other anti-depressive agents for depression
  22 years after approval
  Pooling 115 randomised trials with 26,134 people, paroxetine was neither clearly better nor clearly worse than other antidepressants overall, with some evidence it was less effective than citalopram on early response.
  Worth asking
  The evidence says paroxetine is roughly interchangeable with other SSRIs on effect - did a specific symptom or history point you to it for me?
  Paroxetine versus other anti-depressive agents for depression — Purgato M, et al. (2014)
- Effectiveness of paroxetine in the treatment of acute major depression in adults: a systematic re-examination of…
  16 years after approval
  Re-examining 40 published and unpublished placebo-controlled trials (6,391 adults), paroxetine was not superior to placebo on a combined measure of effectiveness and acceptability, and paroxetine patients more often stopped for side effects (77% higher risk) or experienced suicidal tendencies (odds 2.55 times higher).
  Worth asking
  Unpublished trial data made paroxetine look weaker and riskier than the published record - how do you weigh that when prescribing it today?
  Effectiveness of paroxetine in the treatment of acute major depression in adults: a systematic re-examination of published and unpublished data from randomized trials — Barbui C, et al. (2008)
- Restoring Study 329: efficacy and harms of paroxetine and imipramine in treatment of major depression in adolescence
  23 years after approval
  An independent reanalysis of GlaxoSmithKline's Study 329 (275 adolescents) found paroxetine was no more effective than placebo for adolescent depression and was associated with clinically significant increases in suicidal ideation and behaviour - reversing the original 2001 publication's conclusions.
  Worth asking
  Paroxetine is not recommended for people under 18 - given my age and history, what makes it the right SSRI for me rather than one with a cleaner record?
  Restoring Study 329: efficacy and harms of paroxetine and imipramine in treatment of major depression in adolescence — Le Noury J, et al. (2015)
- Withdrawal from these medications can take months, and NICE says so
  30 years after approval
  NICE guideline NG215 covers safe prescribing and managed withdrawal for five groups of medication: opioids, benzodiazepines, gabapentinoids, Z-drugs and antidepressants. It is the closest thing there is to an official answer on how coming off actually goes.
  It states that withdrawal can be difficult and may take several months or more, that symptoms vary widely in type and severity, that they affect both physical and mental health, and that they can be delayed in onset and can persist. Two recommendations are worth quoting to a prescriber. Do not stop a medicine abruptly except in exceptional medical circumstances. And taper using a slow, stepwise reduction proportionate to the current dose, so the decrements get smaller as the dose gets lower — not a fixed cut each time.
  That last detail is the one most commonly missed. Gabapentinoids are the exception in the guideline and are reduced by a fixed amount at each step.
  Worth asking
  Can we write the taper down, what size are the steps near the end, and how long do I hold at each step before the next reduction.
  Medicines associated with dependence or withdrawal symptoms: safe prescribing and withdrawal management for adults (NG215) — National Institute for Health and Care Excellence (2022)
- In 2019 the Royal College of Psychiatrists changed its position on withdrawal
  27 years after approval
  For years people reporting long, severe antidepressant withdrawal were told it lasted a week or two. In May 2019 the Royal College of Psychiatrists published a position statement conceding the point: while withdrawal symptoms are often mild and self-limiting, there is substantial variation, and for some patients symptoms last much longer and are more severe.
  It went further and asked for changes — that guidelines and patient information recognise the potential for severe and long-lasting withdrawal, that pharmacologically-informed tapering guidance be developed, that discontinuation be tapered at a rate the patient can tolerate over potentially several months, and that clinicians actively work to tell withdrawal apart from relapse.
  One thing it explicitly does not say, and it is worth being accurate about: the College states that from a clinical perspective antidepressant use is not associated with dependence in the addiction sense. Withdrawal and dependence are not the same claim.
  Worth asking
  If I feel bad three weeks after a dose reduction, how will we decide whether that is withdrawal or my depression returning, and what do we do differently in each case.
  Position statement on antidepressants and depression (PS04/19) — Royal College of Psychiatrists (2019)
- Treatment-emergent sexual dysfunction related to antidepressants: a meta-analysis.
  17 years after approval
  When sexual function is actually asked about with a questionnaire rather than waited for as a spontaneous complaint, treatment-emergent sexual dysfunction ranged from 25.8% to 80.3% of patients across antidepressants, all significantly above placebo. In descending order of impact the drugs were sertraline, venlafaxine, citalopram, paroxetine, fluoxetine, imipramine, phenelzine, duloxetine, escitalopram, and fluvoxamine. Agomelatine, amineptine, bupropion, moclobemide, mirtazapine, and nefazodone showed no significant difference from placebo.
  Worth asking
  Where does the antidepressant I'm on sit on the sexual side-effect ranking, and is there a drug with a lower rate that would still treat my condition?
  Treatment-emergent sexual dysfunction related to antidepressants: a meta-analysis. — Serretti A, Chiesa A. (2009)
Step 5
What still is not known
Which questions you might reasonably have has nobody answered yet?
- In a randomised-withdrawal trial the placebo group is being withdrawn from a drug, not left untreated. How much of that 39% is relapse and how much is discontinuation?
- Paroxetine has the most difficult discontinuation profile of the SSRIs. Was that discussed when it was started?
- The 2004 boxed warning on suicidality in under-25s came from a pooled re-analysis twelve years after approval, not from these trials.

The paroxetine legal and safety record

Settled and adjudicated matters only, from primary sources — including the litigation that was decided for the manufacturer, and the cases this drug is verifiably not part of.

Read the paroxetine legal and safety record

Deciding about paroxetine?

Side-effect rates from paroxetine's own label — drug vs placebo, with every caveat the label attaches
How long every medication's approval trials ran — paroxetine beside every other drug in the library, from the same FDA record
12 questions to ask before starting a psychiatric medication — each with the study behind it
Already on it? The 10-question annual review — including the honest case for staying
How long every drug here was tested before approval — one chart, all medications

Open paroxetine (Paxil) in veisund

The app has the full approval journey, the resources behind it, and people working through the same questions.

Browse the evidence library

paroxetine (Paxil)

Before you stop or change your paroxetine dose

what the paroxetine warnings mean

the boxed warning: suicidal thoughts and behaviors, by age, and the pediatric line

the pregnancy warning, in the label’s words, and who warned when

cyp2d6: tamoxifen, two contraindicated antipsychotics, and the other interactions

the discontinuation syndrome, and the taper the trials used

the other warnings, in plain words

the six approved uses, their trials, and the trials that failed

side effects, with the label’s numbers

what the research says, in both directions

questions to ask your prescriber about paroxetine

questions people ask

sources for this section

claims about paroxetine we went and checked

How long the paroxetine trials actually ran

Boxed warning: suicidal thoughts and behaviors

What paroxetine is FDA approved for, and the trials behind it

Side effects of paroxetine in the approval trials

More from the paroxetine label: warnings and precautions

Discontinuation syndrome

Discontinuation for adverse reactions in trials

How many Americans take paroxetine

What people report to the FDA about paroxetine

Most-reported reactions

Drug interactions listed on the paroxetine label

Who pays for paroxetine

Who has been paying for paroxetine, 2019–2024

The paroxetine approval, step by step

What the approval was actually based on

The approval

What was added after it was on the market

What independent research has found since

What still is not known

The paroxetine legal and safety record

Deciding about paroxetine?

Open paroxetine (Paxil) in veisund