Fluoxetine: Independent Evidence on Depression, OCD, Side Effects & Withdrawal

Table of contents

• Evidence summary
• What fluoxetine is
• How it works
• What it is prescribed for
• What works and what does not
• Benefits by claim
• Risks and all side effects
• All interactions
• Who should avoid fluoxetine
• Dosage and how to take it
• Follow the money: who makes it and who funded the evidence
• Related research
• Frequently asked questions
• Sources and funding notes

Evidence summary

This table summarises the main claims about fluoxetine. The most independent evidence comes first.

Independent evidence and credibility scorecard

Independence tier: 1 = regulator or government body; 2 = independent academic or Cochrane review with public funding; 3 = academic work that depends heavily on industry-run trials or has notable author conflicts; 4 = manufacturer-authored or manufacturer-funded. Credibility runs from A (highest) to D.

What fluoxetine is

Fluoxetine is a selective serotonin reuptake inhibitor (SSRI) antidepressant. It is available only on prescription and comes as capsules, tablets (including dispersible tablets) and a liquid (NHS). The best-known brand is Prozac. Eli Lilly and Company, headquartered in Indianapolis, Indiana, discovered and developed it. The US FDA first approved it on 29 December 1987 as a new molecular entity under NDA 018936, held by Eli Lilly (Drugs@FDA, Wong et al., 2005). The same NDA also covers Sarafem. Lilly's 2001 annual report describes Sarafem as a fluoxetine product for premenstrual dysphoric disorder (Lilly 2001 annual report).

Fluoxetine has been generic for a long time. Generic fluoxetine entered the US market in early August 2001 (Lilly 2001 annual report). In the UK it is mostly prescribed as generic capsules, tablets and oral solution made by many licence holders. One example is a 20 mg capsule licensed to Flamingo Pharma (UK) Ltd (UK SmPC). The current US Prozac label is marketed by Lilly USA, LLC (FDA label). It is prescription-only in both the UK and the US.

How it works

The FDA label is candid that "the exact mechanism of PROZAC is unknown". It is presumed to work by blocking the reuptake of serotonin into nerve cells in the brain (FDA label). Fluoxetine blocks serotonin uptake much more potently than noradrenaline uptake. It binds muscarinic, histamine and alpha-1 receptors far less than older tricyclic antidepressants do. The label links those receptors to the anticholinergic, sedating and cardiovascular side effects of tricyclics. The NHS summarises the mechanism as "thought to work by increasing the level of serotonin" (NHS).

The long half-life: fluoxetine's defining feature

Fluoxetine's half-life is 1–3 days after a single dose and 4–6 days with regular use. Its active metabolite, norfluoxetine, has a half-life of 4–16 days. Levels therefore build up over weeks, and steady state is reached at around 4–5 weeks. Even after the last dose, "active drug substance will persist in the body for weeks" (FDA label). This has three practical results:

• Dose changes take weeks to show fully. The label warns that changes in dose "will not be fully reflected in plasma for several weeks".
• Withdrawal is usually gentler. Blood levels fall slowly on their own. NICE notes that people on 20 mg a day can sometimes stop safely after a period of alternate-day dosing (NICE NG222).
• Interactions outlast the prescription. At least 5 weeks must pass after stopping fluoxetine before starting an MAOI antidepressant or thioridazine.

Fluoxetine is mainly broken down by the liver enzyme CYP2D6, and it is also a potent inhibitor of that enzyme. It "may make individuals with normal CYP2D6 metabolic activity resemble a poor metabolizer" (FDA label). This is the root of many of its drug interactions (see All interactions). Peak blood levels come 6–8 hours after a dose. Food may delay absorption by 1–2 hours but does not reduce it.

What it is prescribed for

What works and what does not

Benefits by claim

Adult depression: real, modest, well tolerated

The largest independent synthesis is Cipriani and colleagues' 2018 network meta-analysis. It covered 522 double-blind trials and 116,477 adults, including unpublished data for 52% of the trials. All 21 antidepressants were more effective than placebo, with odds ratios for response ranging from 2.13 (amitriptyline) to 1.37 (reboxetine). The paper's text does not give fluoxetine's individual efficacy odds ratio against placebo, so this article does not quote one. Fluoxetine stood out on acceptability: its dropout odds ratio versus placebo was 0.88 (95% CrI 0.80–0.96), and only agomelatine shared that result. In head-to-head trials, fluoxetine was grouped with fluvoxamine, reboxetine and trazodone as "the least efficacious drugs" (ORs 0.51–0.84), while also being among the more tolerable (Cipriani et al., Lancet 2018).

Two points matter when reading those comparisons. First, the authors found a "novelty effect": a drug looked about 1.18 times more effective when it was the new experimental drug than when it was the older comparator. Fluoxetine, as the long-established reference drug, has usually been the comparator. Adjusting for this "diminished the differences between antidepressants". Second, the certainty of evidence was rated moderate to very low (Cipriani 2018).

The Cochrane review of fluoxetine against other antidepressants (171 studies, 24,868 participants) points the same way. Fluoxetine was as effective as tricyclics as a group (OR 0.97, 95% CI 0.77–1.22) and better tolerated (total dropout OR 0.79; NNT 20). It was less effective than sertraline (OR 1.37; NNT 13), mirtazapine (OR 1.46; NNT 12) and venlafaxine (OR 1.29; NNT 11). The authors cautioned that the "clinical meaning of these differences is uncertain", and that most included studies were sponsored by drug companies (Magni et al., Cochrane 2013).

How big is the benefit over placebo?

Kirsch and colleagues analysed the complete FDA licensing datasets for fluoxetine and three other newer antidepressants, obtained under the US Freedom of Information Act. Drug groups improved by 9.60 points on the Hamilton scale and placebo groups by 7.80. The difference of 1.80 points fell below the 3-point criterion NICE was then using for clinical significance. The gap grew with baseline severity, and the authors say this happened because placebo response dropped in very severe depression, not because the drug worked better (Kirsch et al., PLoS Med 2008). This finding is widely debated. Cipriani's larger 2018 analysis concluded that all antidepressants are more efficacious than placebo, with "mostly modest" effect sizes. In plain terms, many people improve on placebo in trials, and fluoxetine adds a modest average increment on top. Averages can hide people who respond much more, or not at all.

Children and young people: the reference drug, with a small average effect

Fluoxetine's special place in youth depression rests on three lines of evidence:

• Cipriani 2016 (34 trials, 5,260 participants, 14 antidepressants): only fluoxetine was statistically significantly more effective than placebo (SMD −0.51, 95% CrI −0.99 to −0.03). It was also better tolerated than duloxetine and imipramine. The authors concluded that antidepressants "do not seem to offer a clear advantage" for young people overall, but that "fluoxetine is probably the best option to consider when a pharmacological treatment is indicated" (Cipriani et al., Lancet 2016).
• Cochrane 2021 (26 studies): fluoxetine reduced CDRS-R scores by 2.84 points more than placebo (95% CI −4.12 to −1.56; moderate certainty), on a scale that runs from 17 to 113. The authors describe effects of this size as "small and unimportant" on average. They add that "some individuals may experience a greater response" and that sertraline, escitalopram and duloxetine could also be considered as first options (Hetrick et al., Cochrane 2021).
• TADS (NIMH-funded, 439 adolescents): after 12 weeks, response rates were 71.0% with fluoxetine plus CBT, 60.6% with fluoxetine alone, 43.2% with CBT alone and 34.8% with placebo. Clinically significant suicidal thinking, present in 29% at the start, improved in all four groups, most of all with combination treatment. Seven of the 439 participants (1.6%) attempted suicide, and there were no completed suicides (March et al., JAMA 2004).

One analyst argues that many industry-sponsored youth depression trials were "failed trials" with high placebo response. By contrast, the two NIMH-funded trials showed placebo response of 30–35% and between-group differences of 25–30% (Walkup, Am J Psychiatry 2017). This is one expert's interpretation, and funding details for the review were not visible in its abstract. It does help explain why fluoxetine, which has publicly funded trial support, looks better than other antidepressants in young people.

This is why NICE says that when an antidepressant is prescribed for moderate to severe depression in a child or young person, "it should be fluoxetine as this is the only antidepressant for which clinical trial evidence shows that the benefits outweigh the risks". It should be given only with concurrent psychological therapy and after assessment by a child and adolescent psychiatrist (NICE NG134).

OCD

Across 17 trials (3,097 adults), SSRIs lowered YBOCS scores by 3.21 points more than placebo (95% CI −3.84 to −2.57) and nearly doubled the chance of response (RR 1.84). Individual SSRIs did not differ significantly (Soomro et al., Cochrane 2008). In Lilly's two 13-week licensing studies, fluoxetine lowered YBOCS by about 4–9 points, compared with a 1-point drop on placebo (FDA label). NICE advises that response can take up to 12 weeks, and that an effective SSRI should continue for at least 12 months (CG31).

Bulimia nervosa

In the licensing studies, only 60 mg a day was statistically better than placebo. The average reduction versus placebo was 1–2 binge episodes and 2–4 vomiting episodes per week, and the effect did not depend on baseline depression (FDA label). Across 19 antidepressant trials, 5 of them with fluoxetine, the NNT for a 50% cut in binges was 4. Fluoxetine had fewer all-cause dropouts than placebo (Bacaltchuk & Hay, Cochrane 2003). NICE nonetheless positions medication only as an add-on to psychological treatment (NICE NG69).

Risks and all side effects

According to the UK SmPC, the most commonly reported side effects are headache, nausea, insomnia, fatigue and diarrhoea, and they often lessen with continued treatment (SmPC). The NHS lists headaches, nausea and vomiting, diarrhoea or constipation, weight changes, sexual problems, sleep problems, dizziness or drowsiness, and blurred vision as common (NHS). In pooled US placebo-controlled trials, nausea affected 22% on fluoxetine vs 9% on placebo, insomnia 19% vs 10%, anxiety 12% vs 6%, somnolence 12% vs 5%, tremor 9% vs 2% and sweating 7% vs 3% (FDA label).

FDA boxed warning: suicidal thoughts and behaviours

The Prozac label carries a boxed warning. In short-term studies, antidepressants increased suicidal thoughts and behaviour in children, adolescents and young adults. The studies did not show an increase in people over 24, and showed a reduction in people aged 65 and over. The label says to monitor patients of all ages closely, and that Prozac is not approved for children under 7. The pooled FDA data behind it cover 24 paediatric trials (over 4,400 patients) and 295 adult trials (over 77,000 patients). Compared with placebo, the extra cases per 1,000 patients were 14 in under-18s and 5 at ages 18–24. There was 1 fewer case at 25–64 and 6 fewer at 65 and over. "No suicides occurred in any of the pediatric trials" (FDA label).

The FDA's own published paediatric analysis adds a nuance that matters for fluoxetine. Of 24 trials, the publicly funded TADS fluoxetine trial was the only individual trial to show a statistically significant risk ratio for suicidality (4.62, 95% CI 1.02–20.92). The overall risk difference across all drugs was 0.02 (Hammad et al., Arch Gen Psychiatry 2006). Cochrane rates the evidence that fluoxetine may "at least slightly" increase suicide-related outcomes in young people as low certainty (OR 1.27, 95% CI 0.87–1.86) (Hetrick 2021). In adults, the FDA found the risk "strongly age dependent" (Stone et al., BMJ 2009). NICE advises a review 1 week after starting for anyone aged 18 to 25 (NG222). For young people, NICE advises close monitoring, for example weekly contact for the first 4 weeks (NG134).

Withdrawal (discontinuation) symptoms

Stopping antidepressants abruptly, missing doses or taking less than the full dose can cause withdrawal symptoms. These include dizziness, electric-shock sensations, irritability, anxiety, low mood, sleep problems, sweating, nausea and palpitations. NICE also says that "most people stop antidepressants successfully" (NICE NG222). For fluoxetine specifically, the UK SmPC reports an unusual trial finding. Adverse events after stopping occurred in about 60% of people in both the fluoxetine and placebo groups, and 17% vs 12% of these events were severe. Symptoms usually resolve within 2 weeks but can last 2–3 months or more in some people (SmPC). The FDA label notes that fluoxetine's slow decline in blood levels "may minimize the risk of discontinuation symptoms" (FDA label).

Dependence: the FDA label states that fluoxetine "has not been systematically studied" for abuse, tolerance or physical dependence. Premarketing experience did not show drug-seeking behaviour. Withdrawal symptoms are not the same as addiction, but they are real and are a reason to plan any stop with a prescriber.

Overdose

Reported effects of fluoxetine overdose include seizures (which may be delayed), coma, heart-rhythm problems including QT prolongation and cardiac arrest, and serotonin syndrome (FDA label). The label advises prescribing "the smallest quantity of capsules consistent with good patient management". The NHS says to get help from NHS 111 if more than the prescribed dose has been taken, and not to drive yourself to A&E (NHS).

All interactions

Because fluoxetine stays in the body for weeks, interactions can occur for weeks after the last dose (SmPC). Tell every prescriber, dentist and pharmacist that you take it, or took it recently.

Who should avoid fluoxetine

• Contraindications: allergy to fluoxetine; current use of an MAOI or use within 14 days; pimozide or thioridazine; and (UK) metoprolol for heart failure (FDA label, SmPC).
• Needs specialist caution: the NHS says fluoxetine may not be suitable if you have had an allergic or serious reaction to an SSRI, or if you have diabetes, epilepsy, heart disease, glaucoma, a bleeding disorder or a history of mania (NHS). People with long-QT syndrome or recent heart attack need particular care, because such patients were excluded from premarketing trials (FDA label).
• Possible bipolar disorder: screen before starting, because an antidepressant alone may trigger mania.
• Children and teenagers: use only under specialist supervision, for depression, together with psychological therapy. Monitor growth and puberty (NG134, SmPC).
• Pregnancy: the NHS says fluoxetine "can be used during pregnancy if needed", at the lowest effective dose, and usually with a hospital birth so the baby can be monitored (NHS). In 2010 the MHRA reported a possible small increase in congenital heart defects (OR 1.43, 95% CI 0.83–2.47), with absolute risk under 2 in 100 against a background of about 1 in 100 (MHRA). The UK teratology service's Bumps leaflet says more recent studies "generally do not support" a heart-defect link. It estimates persistent pulmonary hypertension of the newborn at around 1 in 300 SSRI-exposed babies and notes a slightly higher risk of postpartum haemorrhage (Bumps/UKTIS). Newborns exposed late in pregnancy can have breathing, feeding and jitteriness problems (FDA label). Stopping can cause relapse, so do not stop without advice.
• Breastfeeding: fluoxetine and norfluoxetine pass into breast milk. Norfluoxetine was found in 85% of sampled infants' blood in one study, and agitation, poor feeding and poor weight gain have been reported (FDA label). The NHS says a doctor may suggest switching to a different antidepressant such as sertraline (NHS).
• Older adults: the 2023 AGS Beers Criteria list SSRIs among drug classes to avoid in people with a history of falls or fractures. They also advise using SSRIs with caution because of SIADH and hyponatraemia, and monitoring sodium closely when starting or changing doses (AGS Beers 2023). The UK SmPC says the daily dose for older people "should generally not exceed 40 mg" (SmPC).
• Liver disease: in cirrhosis, fluoxetine's half-life rose to a mean of 7.6 days, compared with 2–3 days in people without liver disease. A lower or less frequent dose is needed (FDA label).
• Kidney disease: in 12 dialysis patients, blood levels were comparable to those with normal kidneys. The label says a lower dose "is not routinely necessary", but the prescriber will judge (FDA label).

Dosage and how to take it

Your prescriber sets and adjusts your dose. The ranges below are the official licensed ranges, shown for information only. They are not a guide to self-adjustment.

Sources: UK SmPC, FDA label, NICE NG134.

How to take it: usually once a day, at the same time each day, with or without food. Swallow capsules whole. Measure liquid with the syringe or spoon provided (NHS). If you miss a dose, take it when you remember unless the next dose is nearly due. Never take two doses at once.

How long before it works: the NHS says it "can take a few weeks" (NHS). NICE expects benefits in depression within 4 weeks, with a review usually within 2 weeks of starting (NG222). The label says the full effect may take 4 weeks or longer for depression and 5 weeks or longer for OCD. For OCD, NICE says onset can take up to 12 weeks (CG31).

How long to continue: for depression, antidepressants are typically taken for at least 6 months, including after symptoms go (NG222). For young people, NICE advises continuing for at least 6 months after remission (NG134). For OCD, NICE advises at least 12 months (CG31).

How to stop: only with your prescriber. The NHS says the doctor will reduce the dose gradually over several weeks or months (NHS). The UK SmPC says to taper over at least 1–2 weeks (SmPC). NICE notes that because fluoxetine acts for so long, people on 20 mg can sometimes stop using a period of alternate-day dosing. People on 40–60 mg need a gradual schedule, with 1–2 weeks between reductions to judge the effect (NG222). If symptoms are intolerable, the prescriber may return to the previous dose and taper more slowly.

Follow the money: who makes it and who funded the evidence

Originator and commercial history

• Originator: Eli Lilly and Company, headquartered in Indianapolis, Indiana, USA. Fluoxetine was discovered by Lilly scientists in the 1970s, when evidence was emerging that serotonin played a role in depression. It was approved by the FDA in 1987 (Wong et al., 2005, a first-hand account by scientists involved in its development; Drugs@FDA).
• Revenue: Lilly's 2001 annual report to the SEC states that US Prozac sales accounted for "approximately 20 percent" of the company's overall sales in 2000. Worldwide sales of fluoxetine products (Prozac, Prozac Weekly and Sarafem) were $1.99 billion in 2001, down 23% after generic fluoxetine entered the US market in early August 2001 (Lilly 2001 annual report).
• Today: branded Prozac in the US is still marketed by Lilly USA, LLC (FDA label). Most prescriptions worldwide are cheap generics from many manufacturers. Examples are the UK licence holders listed on the electronic medicines compendium, such as Flamingo Pharma (UK) Ltd (SmPC). Because generics dominate, any single company's commercial stake in promoting fluoxetine is probably much smaller than it was in the 1990s.

Who funded the evidence

• Licensing trials: the efficacy and safety trials summarised in the FDA label were submitted under Lilly's New Drug Application (Drugs@FDA). They are manufacturer evidence. The label and FDA reviewers scrutinise them, but they are not independent.
• The wider antidepressant trial base: 409 of 522 trials (78%) in the Cipriani 2018 analysis were funded by pharmaceutical companies. The authors found industry funding "was not associated with substantial differences" in response or dropout, but noted that non-industry trials were few and many trials did not disclose funding (Cipriani 2018). The Cochrane fluoxetine review warned of possible "sponsorship bias" because most studies were drug-company sponsored (Magni 2013).
• Independent syntheses: Cipriani 2018 was publicly funded (UK NIHR; Japan Society for the Promotion of Science), and the funder had no role. Some authors declared personal fees from companies, including lecture fees from Eli Lilly. Cipriani 2016 was funded by China's National Basic Research Program. Hetrick 2021 (Cochrane) was funded by universities and public or charitable bodies in New Zealand and Australia.
• Publicly funded trial: TADS, the key adolescent trial, was sponsored by the US National Institute of Mental Health (ClinicalTrials.gov). It produced both the strongest evidence of benefit and the only individually significant suicidality signal in the FDA's paediatric review (Hammad 2006). Independent research does not always flatter a drug.
• Safety analyses: the suicidality meta-analyses behind the boxed warning were done by FDA staff using data the companies submitted (Hammad 2006, Stone 2009).
• Critical analyses: Kirsch 2008 had no specific funding. Its lead author declared consulting fees from Squibb and Pfizer (Kirsch 2008).

Related research

• Stress, anxiety and depression: evidence-based prevention guide
• Supplements for stress, anxiety and depression: independent evidence
• St John's wort: evidence and interactions (do not combine with fluoxetine)
• Saffron for depression: evidence review
• Omega-3 and vitamin D evidence reviews
• Magnesium, L-theanine and ashwagandha evidence reviews
• Sleep prevention guide and sleep supplements evidence. Insomnia is a common fluoxetine side effect.
• Sibling medicine reviews: sertraline, citalopram, escitalopram, paroxetine, venlafaxine, mirtazapine

Frequently asked questions

Sources and funding notes

• NHS: Fluoxetine (page last reviewed 1 June 2026). UK National Health Service, government-funded; no commercial sponsor.
• NICE NG222: Depression in adults: treatment and management (published 29 June 2022). UK government-funded guideline body.
• NICE NG134: Depression in children and young people (published 25 June 2019). UK government-funded.
• NICE CG31: Obsessive-compulsive disorder and body dysmorphic disorder (published 29 November 2005). UK government-funded.
• NICE NG69: Eating disorders (published 2017, updated 16 December 2020). UK government-funded.
• FDA-approved Prozac prescribing information (DailyMed). Regulator-approved label, marketed by Lilly USA, LLC (US). Efficacy data come from sponsor trials.
• Drugs@FDA: NDA 018936. US regulator record; sponsor Eli Lilly; original approval 29 December 1987.
• UK Summary of Product Characteristics, Fluoxetine 20 mg capsules. Regulator-approved text; licence holder Flamingo Pharma (UK) Ltd (generic manufacturer).
• MHRA Drug Safety Update: SSRIs/SNRIs and postpartum haemorrhage (2021). UK regulator.
• MHRA Drug Safety Update: Fluoxetine and congenital cardiac defects (2010). UK regulator.
• MHRA Drug Safety Update: Tamoxifen and CYP2D6 inhibitors (2010). UK regulator.
• Bumps: Fluoxetine in pregnancy (UKTIS, v3.1, June 2022). Not-for-profit service funded by the UK Health Security Agency.
• Cipriani et al., Lancet 2018: 21 antidepressants network meta-analysis. Funded by NIHR Oxford Health BRC and the Japan Society for the Promotion of Science; funder had no role; 78% of included trials pharma-funded; some authors declared industry fees, including from Eli Lilly (UK/Japan/international).
• Cipriani et al., Lancet 2016: antidepressants in children and adolescents network meta-analysis. Funded by China's National Basic Research Program (973) (UK/China/international).
• Hetrick et al., Cochrane 2021: newer antidepressants for depression in children and adolescents. University, public and charitable funding (NZ/Australia); lead author runs a CBT plus fluoxetine vs placebo trial; one author has a potential interest in a digital therapy.
• Magni et al., Cochrane 2013: fluoxetine versus other pharmacotherapy for depression. Academic Cochrane review; most included trials drug-company sponsored; one author declared Eli Lilly and other honoraria.
• March et al., JAMA 2004: Treatment for Adolescents With Depression Study (TADS). Sponsored by the US National Institute of Mental Health (NCT00006286).
• Hammad et al., Arch Gen Psychiatry 2006: suicidality in paediatric antidepressant trials. FDA staff analysis of company-submitted and NIMH trial data (US).
• Stone et al., BMJ 2009: suicidality in adult antidepressant trials. FDA staff analysis; no competing interests declared (US).
• Kirsch et al., PLoS Medicine 2008: initial severity and antidepressant benefits. No specific funding; lead author declared consulting fees from Squibb and Pfizer (UK-led).
• Soomro et al., Cochrane 2008: SSRIs versus placebo for OCD. Academic (UK-led); no conflicts stated.
• Bacaltchuk & Hay, Cochrane 2003: antidepressants versus placebo for bulimia nervosa. Academic (Brazil-led); no industry funding identified in the abstract.
• Walkup, Am J Psychiatry 2017: industry- and NIMH-funded youth antidepressant trials. Single-author academic review (US); funding and conflicts not visible in the abstract.
• American Geriatrics Society 2023 Beers Criteria. Professional society; "no sponsor for this paper"; individual panel conflicts disclosed (US).
• Wong, Perry & Bymaster, Nat Rev Drug Discov 2005: the discovery of fluoxetine. Historical account by scientists involved in fluoxetine's development at Eli Lilly; used only for history (US).
• Eli Lilly and Company 2001 annual report (SEC filing, 10-K405 exhibit 13). Manufacturer's own financial disclosure; used only for sales and generic-entry facts (US).

Last reviewed: 26 September 2026. Educational information from Pure City Research, not medical advice. Medicine decisions belong with a qualified prescriber.