atomoxetine (Strattera)

the boxed warning, in plain words

the boxed warning reads, in full: "all atomoxetine-treated pediatric patients 6 years of age or older should be monitored and observed closely for suicidal thoughts and behavior, clinical worsening, or unusual changes in behavior, especially during the initial months of therapy or at times of dosage changes. families and caregivers should be advised of the need for close observation and communication with the health care provider. consider stopping atomoxetine in patients who experience emergent suicidal thoughts and behavior. atomoxetine increased the risk of suicidal ideation in pediatric patients aged 6 years of age and older with attention-deficit/hyperactivity disorder (adhd) in short-term studies" [label boxed warning].

the number behind it is in section 5.1. across 12 pooled placebo-controlled studies of 6 to 18 weeks (11 in adhd, one in another condition) with over 2,200 children and adolescents, "the mean incidence of suicidal ideation in atomoxetine-treated pediatric patients was 0.4% (5/1,357) (including one patient with a suicide attempt) compared to 0% (0/851) in placebo-treated pediatric patients. no suicides occurred in these studies." the label adds three facts that shape what the warning means: "all the suicidal ideations occurred in pediatric patients 6 to 12 years of age, and all occurred during the first month", "it is unknown whether the risk of suicidal ideation in pediatric patients extends to longer-term use", and "a similar analysis in adult patients treated with atomoxetine for adhd did not reveal an increased risk of suicidal ideation or behavior" [label 5.1].

so in plain words: five children out of about 1,360 on the drug, none out of about 850 on placebo, all aged 6 to 12, all in the first month, none fatal. it is a small absolute number on a rare event, and with zero cases on placebo the label gives no ratio. what the label asks for is watching, not avoidance: families are to be "alerted about the need to monitor patients daily for the emergence of agitation, irritability, unusual changes in behavior, and mental health-related symptoms, as well as the emergence of suicidal thoughts and behavior, and to report such symptoms immediately", and the label lists anxiety, agitation, panic attacks, insomnia, irritability, hostility, aggressiveness, impulsivity, akathisia, hypomania and mania as symptoms reported with the drug that may be "precursors to emerging suicidality", while saying a causal link "has not been established" [label 5.1].

the other warnings, and what each one asks for

the rest of the label’s warnings, in its own order and words [label 5.2 to 5.11, 4]:

• severe liver injury: "postmarketing reports indicate that atomoxetine can cause severe liver injury". no liver injury was detected in trials of about 6,000 patients, but rare cases followed approval, including liver failure and "a case that resulted in a liver transplant"; most began within 120 days of starting. one patient’s injury came back when the drug was restarted, "providing evidence that atomoxetine likely caused the liver injury". the instruction: test liver function "upon the first symptom or sign of liver dysfunction (e.g., pruritus, dark urine, jaundice, right upper quadrant tenderness, or unexplained 'flu like' symptoms)", and the drug "should be discontinued and not restarted in patients with jaundice or laboratory evidence of liver injury" [label 5.2].
• serious heart problems: sudden death has been reported in children "with structural cardiac abnormalities or other serious heart problems", and sudden deaths, stroke and heart attack in adults, where "the role of atomoxetine" is "also unknown". the label asks for a careful history, including any family history of sudden death or ventricular arrhythmia, and a physical exam before starting, and says it "generally should not be used" in children with known serious structural heart problems, cardiomyopathy or serious arrhythmias [label 5.3].
• blood pressure and heart rate: measure both "at baseline, following atomoxetine dosage increases, and periodically while on therapy". it is contraindicated in severe heart or vascular disorders that would worsen with a rise of about 15 to 20 mmhg or 20 beats a minute [label 4, 5.4]. the label’s table 7 gives how often the trials crossed those lines at any point: a heart rate rise of 20 beats a minute or more in 23% of children on the drug against 12% on placebo, and 22% of adults against 8%; a diastolic rise of 15 mmhg or more in 22% of children against 14% [label table 7]. fainting was reported in 0.8% of children across the programme (46 of 5,596) [label 5.4].
• psychosis and mania: hallucinations, delusional thinking and mania "in patients without a prior history of psychotic illness or mania can be caused by atomoxetine at the recommended dosage". patients are to be screened for a personal or family history of bipolar disorder before starting, because they "may be at increased risk of developing mania or mixed episodes" [label 2.1, 5.5, 5.6].
• aggression or hostility: "there is evidence that atomoxetine may cause the emergence or worsening of aggressive behavior or hostility" [label 5.7].
• allergic reactions, including anaphylaxis and angioedema, "although uncommon" [label 5.8].
• urine flow: in adult trials, urinary retention was 1.7% against 0% on placebo and urinary hesitation 5.6% against 0.5% [label 5.9].
• priapism: "prompt medical attention is required in the event of suspected priapism", defined as an erection lasting more than 4 hours, reported rarely in children and adults [label 5.10].
• growth: "closely monitor growth (e.g., weight, height) in pediatric patients". in short trials children on the drug lost an average 0.4 kg while those on placebo gained 1.5 kg; in long open-label studies weight and height lagged for the first 9 to 12 months, then weight rebounded, and at about 3 years height was 0.4 cm below prediction overall, and 1.2 cm below in children who started before puberty [label 5.11].
• interactions: it is contraindicated with or within 14 days of an maoi, with narrow angle glaucoma, and with pheochromocytoma. paroxetine and fluoxetine, which block the liver enzyme cyp2d6, raised atomoxetine exposure 6 to 8 fold in people with a normal enzyme, and the label asks for a slower titration with any strong cyp2d6 inhibitor [label 4, 2.5, 7, 12.3].

why a gene test is on the label

atomoxetine is broken down mainly by cyp2d6, and the label says to "consider genetic testing to determine the patient’s cyp2d6 metabolizer status" [label 2.5]. people with two non-working copies of the gene, called poor metabolizers, clear the drug far more slowly: its half-life is 21.6 hours against 5.2, it builds up 3.3 fold rather than 1.1 fold, and clearance is about a tenth of normal (0.03 against 0.35 l/hr/kg) [label table 9].

that shows up in side effects. in children, poor metabolizers had more insomnia (11% against 6%), tremor (5% against 1%), depression (7% against 4%) and fainting (3% against 1%); their average heart rate rise was 9.4 beats a minute against 5; 11% of them stopped because of side effects against 6% of others; and after two years their weight gain was 2.4 kg below prediction against 0.2 kg [label 5.4, 5.11, 6.1, table 4]. in adults, poor metabolizers had more dry mouth (35% against 17%), erectile dysfunction (21% against 9%) and urinary retention (6% against 1%) [label table 6]. the label’s response is a longer titration interval, 4 weeks rather than 3 days, if symptoms have not improved and the starting dose is tolerated; the starting, target and maximum doses stay the same [label 2.5].

what the trials measured, in whom, and for how long

the label rests on four short pediatric trials, one long pediatric maintenance trial and two adult trials, all placebo-controlled [label 14]. in children the measure was the clinician-scored adhd rating scale, parent version; in adults it was the clinician-rated conners adult adhd rating scale [label 14.1, 14.2].

the pediatric trials ran 6 to 9 weeks: 297 children aged 8 to 18 on fixed doses for 8 weeks, 171 aged 6 to 16 on one morning dose for 6 weeks, and two 9-week trials of 147 and 144 children aged 7 to 13 that also had a methylphenidate arm [label 14.1]. the drug beat placebo in all four. the fixed-dose trial adds two facts most summaries leave out: the lowest dose, 0.25 mg/kg twice a day, "was not superior to the placebo group", and the highest, 0.9 mg/kg twice a day, "did not provide any additional benefit" over 0.6 mg/kg; the dosing section says "no additional benefit has been demonstrated" above 1.2 mg/kg a day [label 2.3, 14.1]. the two adult trials ran 10 weeks in 280 and 256 adults at a mean of about 48 mg twice a day, and both beat placebo [label 14.2].

the maintenance trial took children aged 6 to 15 who responded during 10 weeks of open-label treatment and randomised 292 to stay on the drug and 124 to placebo; those who stayed well for about 8 months were randomised again, 81 and 82. in both phases those on the drug "experienced significantly longer times to relapse" [label 14.1]. this generic label prints the efficacy results as statements of statistical significance only: it gives no point differences or effect sizes for any of the seven trials, so this page cannot give one from the label. the size of the effect comes from the meta-analysis in the research block.

two safety trials matter for families weighing it against a stimulant: in 148 children with adhd and tourette syndrome or chronic tics, it "did not worsen tics" over 18 weeks, though 70% left the trial, mostly for lack of effect on adhd; and in children and adults with adhd and anxiety disorders it did not worsen anxiety [label 14.3].

side effects, with the label’s numbers

in the short pediatric trials, 1,597 children on the drug against 934 on placebo: headache 19% against 15%, abdominal pain 18% against 10%, decreased appetite 16% against 4%, somnolence 11% against 4%, vomiting 11% against 6%, nausea 10% against 5%, fatigue 8% against 3%, irritability 6% against 3%, dizziness 5% against 2%, and weight loss 3% against 0% [label table 2]. 3% of children stopped because of a side effect against 1.4% on placebo [label 6.1]. in the fixed-dose trial, 29.1% of children at the top dose lost at least 3.5% of their body weight, against 1.3% on placebo [label 5.11].

adults report a different pattern. in 1,697 adults on the drug against 1,560 on placebo: nausea 26% against 6%, dry mouth 20% against 5%, decreased appetite 16% against 3%, insomnia 15% against 8%, fatigue 10% against 6%, erectile dysfunction 8% against 1% (of men), constipation 8% against 3%, dizziness 8% against 3%, urinary hesitation 6% against 1%, and delayed ejaculation 4% against 1% [label table 5]. 11.3% of adults stopped because of a side effect against 3% on placebo [label 6.1]. the label says sexual side effects are "likely to underestimate their actual incidence" because patients and clinicians may not raise them, and asks prescribers to "routinely inquire about sexual dysfunction" [label 6.1].

after approval, reports added qt prolongation, raynaud’s phenomenon, seizures (including in people with no history or risk factors), tics, depression and anxiety [label 6.2]. no death has been reported from an overdose of atomoxetine alone, including intentional overdoses up to 1,400 mg; deaths have been reported in mixed overdoses with other drugs [label 10].

what the research says, in both directions

a limit to state first: this site’s library holds no row tagged atomoxetine yet. the research below is the label’s own trials and the cross-drug studies in the library that include it, and the library rows those studies sit on are written up for other adhd drugs.

in favour. the largest network meta-analysis of adhd medicines, 133 double-blind trials, found that on clinician ratings at about 12 weeks atomoxetine beat placebo in children and adolescents (standardised mean difference -0.56, 95% ci -0.66 to -0.45) and in adults (-0.45, -0.58 to -0.32) [cortese-2018]. it is the one adhd drug in a 2024 pharmacovigilance study of people aged 13 to 25 for which no association with reports of psychosis was found, where amphetamines were reported for psychosis more often than methylphenidate; that study counts reports, not rates [hamard-2024]. its label shows no abuse potential and no withdrawal syndrome, and trials found it did not worsen tics or anxiety [label 9.2, 9.3, 14.3].

against, or smaller than it sounds. in the same meta-analysis the effect in children was smaller than for amphetamines (-1.02) or methylphenidate (-0.78), head-to-head comparisons favoured amphetamines over atomoxetine in both age groups, teacher ratings showed a benefit over placebo only for methylphenidate and modafinil among the comparisons available, and adults were more likely to stop atomoxetine because of side effects than placebo (odds ratio 2.33, 95% ci 1.28 to 4.25); the authors recommended methylphenidate first for children and amphetamines first for adults, and found too little data to judge any drug at 26 or 52 weeks [cortese-2018]. a swedish study of 278,027 people with adhd linked each added year of adhd medicine use to a 4% higher risk of cardiovascular disease, mostly high blood pressure and arterial disease; that is a class finding across adhd medicines, written up in the library for methylphenidate, not a result for atomoxetine alone [zhang-2024]. the label’s own pediatric trials were 6 to 9 weeks long, and it reports no trial in people over 65 [label 8.5, 14].

for contrast, the stimulants carry a boxed warning on abuse, misuse and addiction that the fda updated across the class in 2023 [stimulant-warning]; atomoxetine’s boxed warning is a different one, about suicidal thoughts in children, and guanfacine, the other common non-stimulant, carries none [guanfacine-explainer; compare-guanfacine]. the side-by-side pages set the labels next to each other [compare-guanfacine; compare-methylphenidate].

questions people ask

does atomoxetine (strattera) have a black box warning?

yes, one: suicidal thoughts and behaviors in children and adolescents aged 6 and over. in 12 pooled short trials, suicidal ideation was reported in 0.4% of children on atomoxetine (5 of 1,357, including one suicide attempt) against 0% on placebo (0 of 851), with no suicides; all cases were in 6 to 12 year olds and in the first month. a similar analysis in adults found no increased risk. the label asks for close observation, especially early and after dose changes.

how long does atomoxetine take to work?

the label does not give a time to effect. its pediatric trials measured results after 6 to 9 weeks and its adult trials after 10 weeks. the dose is raised after a minimum of 3 days to the target, and in adults the label allows a move to the maximum only if response is not optimal after a further 2 to 4 weeks. in cyp2d6 poor metabolizers or with a strong cyp2d6 inhibitor, it recommends a 4-week titration interval. how long to wait before judging it is a question for the prescriber.

can atomoxetine cause liver damage?

rarely. no liver injury was seen in trials of about 6,000 patients, but after approval the label records rare cases of severe liver injury, including liver failure and one liver transplant, mostly within 120 days of starting. one case recurred on restarting, which the label calls evidence that atomoxetine likely caused it. the label asks for liver tests at the first sign, such as itching, dark urine, jaundice or unexplained flu-like symptoms, and says the drug should not be restarted after jaundice or lab evidence of liver injury.

is atomoxetine a stimulant or a controlled substance?

no to both. the label calls it a selective norepinephrine reuptake inhibitor, says it is not a controlled substance, and reports that in an adult abuse study it had no stimulant or euphoriant properties. after stopping, trials found no rebound or withdrawal syndrome, and the label says no taper is needed. its warnings are about suicidal thoughts in children, the liver, the heart and blood pressure rather than misuse.

what are the most common atomoxetine side effects?

in children: headache, abdominal pain (18% against 10% on placebo), decreased appetite (16% against 4%), sleepiness (11% against 4%), vomiting, nausea, fatigue and irritability. in adults: nausea (26% against 6%), dry mouth (20% against 5%), decreased appetite, insomnia, constipation, dizziness, erectile dysfunction (8% against 1% of men) and urinary hesitation. people who are cyp2d6 poor metabolizers had several of these more often.

how does atomoxetine compare with stimulants for adhd?

a 2018 network meta-analysis of 133 trials found atomoxetine better than placebo on clinician ratings in children and adults, but with a smaller effect than amphetamines or methylphenidate in children, and head-to-head comparisons favoured amphetamines. adults were more likely to stop it than placebo because of side effects. it is not a controlled substance and its label shows no abuse potential, while the stimulant labels carry a class boxed warning on abuse, misuse and addiction.

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.

1. atomoxetine capsules prescribing information, fda label on dailymed (set id 6f6bea0c-297b-43ff-8960-ecc11243e06e, version 17, effective august 14, 2026; a generic of strattera): boxed warning, sections 1, 2.1 to 2.8, 4, 5.1 to 5.11, 6.1, 6.2, 7, 8.5, 9.1 to 9.3, 10, 12.3, 14.1 to 14.3 and tables 2 to 9
2. drugs@fda, nda 021411 (strattera, atomoxetine), approved november 26, 2002
3. ahrq medical expenditure panel survey, prescribed medicines public-use files 2014 (hc-168a) to 2024 (hc-254a): people with at least one atomoxetine purchase, weighted; no reportable year, with 14 to 44 respondents a year against ahrq’s minimum of 60; computed by veisund, method on the usage section of this page
4. cms medicare part d prescribers by geography and drug, national rows, data year 2024: 66,109 beneficiaries across brand and generic rows, summed, an upper bound on distinct people; denominator cms medicare monthly enrollment 2024, 54,375,198 beneficiaries with drug coverage
5. cortese s, et al. comparative efficacy and tolerability of medications for attention-deficit hyperactivity disorder in children, adolescents, and adults: a systematic review and network meta-analysis. lancet psychiatry 2018 (the library row is written up for methylphenidate; the atomoxetine findings were read from the pubmed abstract, pmid 30097390, on 2026-10-13) (library row)
6. hamard j, rousseau v, durrieu g, et al. psychosis with use of amphetamine drugs, methylphenidate and atomoxetine in adolescent and adults. bmj mental health 2024 (a who vigibase pharmacovigilance study of reports, not rates; the library row is written up for lisdexamfetamine; the atomoxetine finding was read from the abstract via crossref, doi 10.1136/bmjment-2023-300876, on 2026-10-13) (library row)
7. zhang l, et al. attention-deficit/hyperactivity disorder medications and long-term risk of cardiovascular diseases. jama psychiatry 2024 (a class finding across adhd medicines; the library row is written up for methylphenidate) (library row)
8. fda drug safety communication, 2023: fda updating warnings to improve safe use of prescription stimulants used to treat adhd and other conditions (stimulants only; cited here for contrast) (library row)
9. what the guanfacine (intuniv) warnings mean, the companion explainer on this site for the other common non-stimulant
10. atomoxetine vs guanfacine, this site’s side-by-side comparison of the two non-stimulant adhd labels
11. atomoxetine vs methylphenidate, this site’s side-by-side comparison

Most-reported reactions

• Drug ineffective2,955
• Nausea2,251
• Fatigue2,246
• Headache1,843
• Vomiting1,815
• Off label use1,799
• Dizziness1,708
• Insomnia1,618
• Depression1,478
• Somnolence1,461