trazodone (Desyrel)

the boxed warning: suicidal thoughts and behaviors, by age

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 that trazodone "is not approved for use in pediatric patients" [label, boxed warning and 5.1]. the number behind it is pooled placebo-controlled trials of antidepressants covering about 77,000 adults and more than 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 1]. 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 warning is specific, not blanket: thoughts and behaviors, not deaths; the first months, not years; and above age 24 the pooled trials did not show an increase at all [label 5.1; age-ceiling]. the pediatric line is a plain statement of absence: "safety and effectiveness of raldesy in the pediatric patients have not been established" [label 8.4]. the warning is written for depression trials at antidepressant doses. the label does not say whether it applies at the low doses used for sleep, and this page does not either.

sleep is not an approved use, and what the trials of that use found

the label’s indications section names one condition, major depressive disorder in adults [label 1]. insomnia is not an approved use, and no sleep trial appears anywhere in the label. the only sleep-adjacent sentence is in the dosing section: "occurrence of drowsiness may require the administration of a major portion of the daily dose at bedtime or a reduction of dosage" [label 2.1]. the label also says to take it shortly after a meal or light snack, and that a high-fat meal lowers the peak blood level by 31% [label 2.1, 12.3].

in favour. the cochrane review of antidepressants for insomnia pooled three trials with 370 participants and found low-dose trazodone gave a moderate short-term improvement in how people rated their sleep against placebo (standardised mean difference -0.34, moderate-certainty evidence) [everitt-2018]. a 2024 meta-analysis of 44 randomised trials in 3,935 people found trazodone improved rated sleep quality and added roughly 28 minutes of sleep-lab-measured sleep against placebo, though at a mean dose of about 179 mg a day, which is an antidepressant dose rather than a low one [kokkali-2024]. a 2018 meta-analysis of seven placebo-controlled trials in 429 people found fewer early-morning awakenings and better rated sleep quality (standardised mean difference -0.41) with good short-term tolerability [yi-2018]. in moderate-to-severe alzheimer’s disease, one 30-person trial of 50 mg for two weeks found about 42 more minutes of night-time sleep and sleep efficiency up about 8.5 percentage points, with no serious adverse effects reported [mccleery-2020].

against, or smaller than it sounds. in the same cochrane review, the two trials that measured sleep in a laboratory rather than by questionnaire found little or no difference in actual sleep efficiency; the review found no evidence on long-term use and rated the adverse-event evidence low certainty [everitt-2018]. in the 44-trial meta-analysis, trazodone did not change how long people felt they slept, more than twice as many people stopped because of side effects (relative risk 2.30), and sleep-related adverse effects were about four times as frequent as on placebo (relative risk 4.31); its funding and conflict-of-interest statements sat behind a paywall and could not be read [kokkali-2024]. the seven-trial meta-analysis found no significant improvement in sleep efficiency or most objective measures [yi-2018]. the alzheimer’s evidence is one small two-week trial rated low certainty, with negligible effects on night-time awakenings and daytime sleep, and longer-term safety in dementia untested [mccleery-2020]. the question none of these trials answers is what happens past a few weeks, which is how long most people take it.

this site has two other pages on the sleep question and does not repeat them here: the evidence page, with the trials, the cochrane review and the sleep guideline [sleep-topic], and the earlier article on what the label does and does not say [sleep-blog]. zolpidem, the drug trazodone is most often compared with for sleep, has its own explainer [zolpidem-explainer].

the heart-rhythm warning, and the dose it names

warning 5.3, cardiac arrhythmias, opens with the label’s own trial finding: "clinical studies indicate that trazodone hydrochloride may be arrhythmogenic in patients with preexisting cardiac disease". the arrhythmias it names are isolated premature ventricular contractions, ventricular couplets, tachycardia with syncope, and torsade de pointes. then the dose line: "postmarketing reports, including torsade de pointes have been reported at doses of 100 mg or less with the immediate-release trazodone hydrochloride tablets" [label 5.3]. the post-marketing section repeats it: "prolonged qt interval, torsade de pointes, and ventricular tachycardia have been reported at doses of 100 mg per day or less" [label 6.2]. that is the dose range in which the sleep trials above were run, and the label says so without saying anything about sleep.

the label states flatly that "trazodone hydrochloride prolongs the qt/qtc interval" and says to avoid it in anyone with known qt prolongation, a history of arrhythmias, symptomatic slow heart rate, low potassium or magnesium, or congenital long qt, and not to use it during the early recovery phase after a heart attack [label 5.3]. it says to avoid combining it with drugs that block the liver enzyme cyp3a4 (it names itraconazole, clarithromycin and voriconazole) or that also prolong the qt interval: class 1a antiarrhythmics such as quinidine and procainamide, class 3 such as amiodarone and sotalol, antipsychotics such as ziprasidone, chlorpromazine and thioridazine, and antibiotics such as gatifloxacin [label 5.3, 7]. a strong cyp3a4 inhibitor raises trazodone exposure and "the risk of adverse reactions, including cardiac arrhythmias, may be increased", so the label says to consider a lower dose [label 7, table 3]. in the trials, syncope was 5% against 1% in outpatients, and "occasional sinus bradycardia has occurred in long-term studies" [label 6.1, table 2]. in overdose, qt prolongation is among the most severe reactions seen with trazodone alone [label 10]. the label gives no rate for torsade and no qt measurement in milliseconds; a post-marketing report says something has happened, not how often.

priapism: the four-hour rule

warning 5.6 says cases of priapism, painful erections lasting more than six hours, have been reported in males taking trazodone tablets, and that untreated priapism "can result in irreversible damage to the erectile tissue". its instruction is unusually direct: "males who have an erection lasting greater than 4 hours, whether painful or not, should immediately discontinue the drug and seek emergency medical attention" [label 5.6]. that is one of the two places on this page where the label itself tells a patient to stop; the other is the serotonin syndrome warning, and both describe emergencies, not routine decisions. the label asks for caution in men with sickle cell anaemia, multiple myeloma, leukaemia, or anatomical deformation of the penis [label 5.6]. the post-marketing list also carries clitorism, the equivalent in women, alongside breast enlargement and lactation [label 6.2]. priapism is also named first among the most severe reactions in overdose with trazodone alone [label 10]. the label gives no frequency: "impotence", "increased libido" and "retrograde ejaculation" appear among reactions seen in fewer than 2% of trial patients, and priapism is a post-marketing report with no denominator [label 6.1, 6.2].

low blood pressure, drowsiness and driving

warning 5.4 says hypotension, including orthostatic hypotension (the drop on standing) and syncope, has been reported, and that someone also on a blood-pressure drug may need that drug’s dose reduced [label 5.4]. the trial table bears it out in two directions: among outpatients, hypotension 4% against 0% and syncope 5% against 1%; among inpatients, hypertension 20% against 1% [label 6.1, table 2]. the label offers no explanation for the inpatient hypertension figure, and this page does not invent one. the pharmacology section says trazodone blocks alpha-1 adrenergic receptors, "a property which may be associated with postural hypotension" [label 12.2].

warning 5.9 says trazodone "may cause somnolence or sedation and may impair the mental and/or physical ability required for the performance of potentially hazardous tasks", and tells patients not to drive or operate machinery until they know how it affects them [label 5.9]. drowsiness was the most common reaction in the trials: 41% against 20% in outpatients and 24% against 6% in inpatients, with dizziness or light-headedness 28% against 15% and 20% against 5% [label 6.1, table 2]. the label adds that trazodone may enhance the effect of alcohol, barbiturates and other central nervous system depressants [label 7], and that deaths from overdose have occurred when trazodone was taken with them, naming alcohol, alcohol with chloral hydrate and diazepam, amobarbital, chlordiazepoxide and meprobamate [label 10]. the elimination half-life is about 18 hours [label 12.3], which the label states as a pharmacokinetic fact without drawing a conclusion about next-day effects; the 44-trial meta-analysis’s sleep-related adverse-effect figure, four times placebo, is the nearest measurement on this page [kokkali-2024].

the other warnings, in plain words

the warnings section runs from 5.1 to 5.11. the label describes trazodone as both a serotonin reuptake inhibitor and a 5-ht2 receptor blocker, and says the net result of that combination "is unknown" [label 12.1], which is why it carries the serotonergic antidepressant class warnings. beyond those above:

• serotonin syndrome [5.2]: too much serotonin signalling, with agitation, hallucinations or delirium, racing heart, swinging blood pressure, sweating, fever, tremor, rigidity, muscle jerking, overactive reflexes, seizures, nausea and diarrhoea. the risk rises with triptans, tricyclics, fentanyl, lithium, tramadol, tryptophan, buspirone and st john’s wort, and it can occur on trazodone alone. use with an maoi, including the antibiotic linezolid and intravenous methylene blue, is contraindicated within 14 days either way [4, 2.3]. the label says to stop trazodone and any other serotonergic drug immediately if the symptoms occur.
• bleeding [5.5]: drugs that interfere with serotonin reuptake raise the risk of bleeding, from bruising and nosebleeds to life-threatening haemorrhage, and case-control and cohort studies have linked them to gastrointestinal bleeding. aspirin, nsaids, other antiplatelet drugs and anticoagulants add to the risk; on warfarin the label asks for the inr to be monitored carefully when trazodone starts, changes or stops [7, table 3].
• mania [5.7]: treating a depressive episode in someone with bipolar disorder can precipitate a mixed or manic episode; the label asks for screening for a personal or family history of bipolar disorder, mania or hypomania first [2.2]. hypomania appears among reactions seen in fewer than 2% of trial patients [6.1].
• glaucoma [5.10]: pupil dilation can trigger an angle-closure attack in an eye with untreated narrow angles; the label says to avoid antidepressants in that situation.
• low sodium [5.11]: hyponatremia, with cases below 110 mmol/l, usually from inappropriate antidiuretic hormone secretion. signs are headache, poor concentration, memory impairment, confusion, weakness and unsteadiness, "which can lead to falls"; severe cases have included hallucination, syncope, seizure, coma, respiratory arrest and death. older people, people on diuretics and people who are volume-depleted are at greater risk [8.5].
• digoxin and phenytoin [7]: both are narrow-margin drugs whose blood levels trazodone can raise; the label says to measure them before starting and to keep monitoring. a strong cyp3a4 inducer lowers trazodone exposure and may call for a higher dose [2.4].
• pregnancy [8.1]: "published prospective cohort studies, case series, and case reports over several decades" have not identified drug-associated risks of major birth defects, miscarriage or other adverse outcomes, with the caveat that all of those studies have methodological limitations, including small samples and inconsistent comparators. the label sets against that the 201-woman study in which stopping antidepressants in pregnancy raised relapse. trazodone transfers into breast milk, and limited post-marketing data have not linked it to harm in the breastfed child [8.2].
• kidney and liver [8.6, 8.7]: trazodone "has not been studied" in people with impaired kidney or liver function; the label asks for caution in both.
• not a controlled substance [9.1, 9.2]: trazodone is not scheduled, has not been systematically studied for abuse potential, and showed no drug-seeking behaviour in its clinical studies.
• post-marketing reports [6.2]: besides the arrhythmias and priapism above, the list includes congestive heart failure, conduction block, atrial fibrillation, myocardial infarction, cardiac arrest, "unexplained death", hemolytic anemia, liver enzyme changes and jaundice, grand mal seizures, stroke, extrapyramidal symptoms and tardive dyskinesia, psychosis and paranoid reaction, urinary retention and incontinence, apnea, and hair loss. these are voluntary reports from a population of uncertain size, and the label says their frequency and cause cannot be established from them.

what the label says about its own trials, which is almost nothing

the clinical studies section is one sentence: "the efficacy of raldesy for the treatment of mdd in adults is based on studies of trazodone hydrochloride tablets" [label 14]. the label does not say how many trials there were, how long they ran, what scale they used or what they showed. the only trace of them is the adverse-reaction table, which indicates the controlled studies included 142 inpatients and 157 outpatients on trazodone against 95 and 158 on placebo, with durations not stated [label 6.1, table 2]. the fda has not published its 1981 review, and the approval record on drugs@fda gives the date and the application number [approval]. for a drug 6.4 million americans take, the label’s own efficacy record is thinner than any other on this site.

the dose it describes for depression is 150 mg a day in divided doses, rising by 50 mg every three or four days to a usual maximum of 400 mg for outpatients and 600 mg for inpatients [label 2.1]. the research section below has what independent trials found when trazodone was measured against other antidepressants at those doses.

side effects, with the label’s numbers

the label’s one-line summary: the most common reactions, at 5% or more and twice the placebo rate, are "edema, blurred vision, syncope, drowsiness, fatigue, diarrhea, nasal congestion, weight loss" [label 6]. among the 157 outpatients on trazodone against 158 on placebo: drowsiness 41% against 20%, dry mouth 34% against 20%, dizziness or light-headedness 28% against 15%, headache 20% against 16%, blurred vision 15% against 4%, nausea or vomiting 13% against 10%, constipation 8% against 6%, allergic skin condition or edema 7% against 1%, fatigue 6% against 3%, nasal or sinus congestion 6% against 3%, weight loss 6% against 3%, diarrhea 5% against 1%, syncope 5% against 1%, weight gain 5% against 2%, tremors 5% against 4%, aches and pains 5% against 3%, hypotension 4% against 0% [label 6.1, table 2]. confusion ran the other way in outpatients, 6% against 8%, and nervousness 6% against 8% [label 6.1, table 2].

among the 142 inpatients against 95: drowsiness 24% against 6%, dizziness 20% against 5%, hypertension 20% against 1%, dry mouth 15% against 8%, nervousness 15% against 11%, fatigue 11% against 4%, headache 10% against 5%, nausea or vomiting 10% against 1%, hypotension 7% against 1%, constipation 7% against 4%, confusion 5% against 0%, incoordination 5% against 0% [label 6.1, table 2]. reactions in fewer than 2% included akathisia, hallucinations or delusions, hypomania, impaired memory and speech, impotence, increased libido, retrograde ejaculation, early or missed periods, muscle twitches, numbness, shortness of breath and palpitations [label 6.1]. the label gives no dropout rate for side effects and no figures by dose. in overdose the most frequent reactions have been drowsiness and vomiting, and there is no specific antidote [label 10].

the discontinuation syndrome

warning 5.8 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. "a gradual reduction in dosage rather than abrupt cessation is recommended whenever possible" [label 5.8, 2.5]. unlike some antidepressant labels, this one gives no taper schedule, no step size and no duration. it also does not say whether the syndrome occurs at the low doses used for sleep. nothing in the library measures that for trazodone either, so this page leaves the question open and lists it below as one to ask.

what the research says about the approved use, in both directions

in favour. in the network meta-analysis of 522 trials covering 116,477 adults with major depression, trazodone worked better than placebo [cipriani-2018]. that analysis covered trazodone at antidepressant doses for depression, which is what the label approves; it was publicly funded, and several of its authors disclosed pharmaceutical lecture and consulting fees [cipriani-2018].

against, or smaller than it sounds. in the same analysis’s head-to-head comparisons, trazodone was among the least effective of the 21 antidepressants and had among the highest dropout rates, with odds ratios of 1.30 to 2.32 for stopping treatment against other drugs [cipriani-2018]. the label’s own efficacy section, one sentence long, offers nothing to set against that [label 14]. the sleep evidence, both ways, is in the block above; the summary is that the approved use has a thin public record and the common use has a short one.

questions people ask

what does the trazodone (desyrel) black box warning mean?

it is the antidepressant class warning. in pooled placebo-controlled trials of antidepressants covering about 77,000 adults and more than 4,500 children, 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 box also says trazodone is not approved for anyone under 18, and the label says its safety and effectiveness in children have not been established. the warning says to watch everyone starting an antidepressant, especially in the first months and after dose changes. it describes thoughts and behaviors, not deaths, and it was written from depression trials at antidepressant doses; the label does not say whether it applies at the low doses used for sleep.

is trazodone approved for sleep?

no. the label approves trazodone for one condition, major depressive disorder in adults, at 150 to 400 mg a day. insomnia is not an approved use, and no sleep trial appears anywhere on the label; its only sleep-adjacent line says drowsiness may call for taking most of the daily dose at bedtime. the cochrane review of antidepressants for insomnia calls trazodone the most commonly prescribed antidepressant for insomnia in the united states and says that use is off-label. off-label prescribing is legal and common; what it means here is that the fda has never reviewed evidence that trazodone works for sleep, and the evidence that exists is summarised in the next answer.

does trazodone work for sleep?

a little, for a few weeks, by questionnaire more than by measurement. the cochrane review pooled three trials in 370 people and found low-dose trazodone improved how people rated their sleep against placebo, a moderate effect with moderate-certainty evidence, but the two trials with sleep-lab measurement found little or no difference in actual sleep efficiency, and there was no evidence on long-term use. a 2024 meta-analysis of 44 trials in 3,935 people found about 28 more minutes of measured sleep and better rated quality at a mean dose of about 179 mg, with more than twice the dropout for side effects and sleep-related adverse effects about four times as frequent as placebo. a seven-trial meta-analysis found fewer early-morning awakenings and better rated quality but no gain in sleep efficiency. none of these trials ran long enough to answer the question most people taking it for years would ask. this site’s evidence page on the sleep question has the detail.

does trazodone cause heart problems or qt prolongation?

the label says it can. warning 5.3 states that "trazodone hydrochloride prolongs the qt/qtc interval", that its trials indicated it may cause arrhythmias in people with pre-existing heart disease, and that torsade de pointes, qt prolongation and ventricular tachycardia have been reported after approval "at doses of 100 mg per day or less", which is the range used for sleep. the label says to avoid it in anyone with known qt prolongation, a history of arrhythmias, a slow heart rate, low potassium or magnesium or congenital long qt, not to use it in early recovery from a heart attack, and to avoid combining it with other qt-prolonging drugs or with strong cyp3a4 blockers such as clarithromycin, itraconazole and voriconazole. it gives no rate: post-marketing reports show that something has happened, not how often. in the trials, fainting was 5% against 1%.

what is the trazodone priapism warning?

priapism is a prolonged erection, and the label says cases lasting more than six hours have been reported in men taking trazodone and that untreated it can permanently damage erectile tissue. the label’s instruction is that a man with an erection lasting more than four hours, painful or not, should immediately stop the drug and seek emergency care. it asks for caution in men with sickle cell anaemia, multiple myeloma, leukaemia or a penile deformity, and the post-marketing list includes clitorism, the equivalent in women. the label gives no frequency; priapism is a post-marketing report without a denominator, and it is also named first among the most severe reactions in overdose.

what are the most common trazodone side effects?

in the label’s outpatient trials, 157 people on trazodone against 158 on placebo: drowsiness 41% vs 20%, dry mouth 34% vs 20%, dizziness or light-headedness 28% vs 15%, headache 20% vs 16%, blurred vision 15% vs 4%, nausea or vomiting 13% vs 10%, constipation 8% vs 6%, allergic skin condition or edema 7% vs 1%, fatigue, nasal congestion and weight loss 6% vs 3%, diarrhea and syncope 5% vs 1%, weight gain 5% vs 2%, hypotension 4% vs 0%. among inpatients, hypertension was 20% vs 1%, which the label reports without explaining. the label’s own summary of reactions at 5% or more and twice placebo is edema, blurred vision, syncope, drowsiness, fatigue, diarrhea, nasal congestion and weight loss. it gives no dropout rate and no figures by dose; the independent trials of the sleep use found dropout for side effects more than twice placebo.

what happens if you stop trazodone suddenly?

the label lists the serotonergic antidepressant discontinuation syndrome: nausea, sweating, low mood, irritability, agitation, dizziness, electric-shock sensations, tremor, anxiety, confusion, headache, lethargy, emotional lability, insomnia, hypomania, tinnitus and seizures, particularly after abrupt stopping, and it recommends a gradual reduction whenever possible. it gives no schedule and does not say whether the syndrome occurs at the low doses used for sleep; nothing in the library measures that either. the two situations in which the label itself tells a patient to stop immediately are an erection lasting more than four hours and the symptoms of serotonin syndrome, both as emergencies. anything else is a prescriber conversation.

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. trazodone hydrochloride oral solution (raldesy) prescribing information, fda label on dailymed (set id 34708cc1-b0b2-4fd0-b9f3-7f008b457fc9, version 6, effective september 1, 2026; validus, nda 218637): boxed warning; sections 1, 2.1 to 2.5, 4, 5.1 to 5.11 with table 1, 6, 6.1 table 2, 6.2, 7 table 3, 8.1, 8.2, 8.4 to 8.7, 9.1, 9.2, 10, 12.1 to 12.3, 14
2. drugs@fda, nda 018207 (desyrel, trazodone hydrochloride tablets), approved december 24, 1981
3. ahrq medical expenditure panel survey, prescribed medicines public-use files 2014 (hc-168a) and 2024 (hc-254a): people with at least one trazodone purchase and prescriptions, weighted; the 2014 to 2024 prescription change is within the survey’s margin of error; computed by veisund, method on the usage section of this page
4. the antidepressant suicidality warning’s age ceiling, from the labels themselves, as a library row (library row)
5. everitt h, et al. antidepressants for insomnia in adults. cochrane database of systematic reviews 2018 (three trazodone trials, 370 participants) (library row)
6. kokkali m, et al. effects of trazodone on sleep: a systematic review and meta-analysis. cns drugs 2024 (44 randomised trials, 3,935 participants) (library row)
7. yi xy, et al. trazodone for the treatment of insomnia: a meta-analysis of randomized placebo-controlled trials. sleep medicine 2018 (seven trials, 429 participants) (library row)
8. mccleery j, sharpley al. pharmacotherapies for sleep disturbances in dementia. cochrane database of systematic reviews 2020 (one trazodone trial, 30 participants, two weeks) (library row)
9. cipriani a, et al. 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. lancet 2018 (library row)
10. does trazodone work for sleep? the trials, the cochrane review and the sleep guideline, a separate evidence page on this site
11. trazodone for sleep: what the label actually says, an earlier article on this site
12. what the zolpidem (ambien) warnings mean, the companion explainer on this site for the approved sleep drug trazodone is most often compared with

Most-reported reactions

• Drug ineffective1,212
• Fatigue1,157
• Nausea1,124
• Headache924
• Pain823
• Diarrhoea817
• Off label use742
• Vomiting707
• Insomnia695
• Depression674