The serotonin story: what the evidence said, and what patients were told

Almost everyone who has been prescribed an antidepressant in the last thirty years has heard some version of this: depression is caused by a chemical imbalance, the imbalance involves serotonin, and the medication corrects it. Often it was said kindly, by someone trying to lift a weight of self-blame off a patient's shoulders. Often it worked, as reassurance.

The problem is what happened to the sentence underneath. The research it rested on moved on. The sentence did not.

This is a piece about the gap between those two things. It is not a piece arguing that antidepressants do not work — and the difference between those two claims is the most important thing on this page.

This is not medical advice and it is not a reason to change anything. Stopping antidepressants abruptly can be dangerous. If you are in crisis, call or text 988 (US), 24/7, free.

What the umbrella review found

In 2022, a team led by Joanna Moncrieff published a systematic umbrella review in Molecular Psychiatry — a review of the systematic reviews and meta-analyses across the main areas of serotonin research [1]. It was registered in advance with PROSPERO, and it covered six research areas: serotonin and 5-HIAA concentrations in body fluids; 5-HT1A receptor binding; serotonin transporter (SERT) levels; tryptophan depletion studies; SERT gene associations; and gene-environment interactions.

Seventeen studies were included. The conclusion, verbatim [1]:

"The main areas of serotonin research provide no consistent evidence of there being an association between serotonin and depression, and no support for the hypothesis that depression is caused by lowered serotonin activity or concentrations."

Some of the individual findings are worth reading, because they show the size of the studies involved rather than asking you to take a summary on trust [1]:

• Two meta-analyses of the serotonin metabolite 5-HIAA showed no association with depression (largest n = 1,002).
• A meta-analysis of cohort studies of plasma serotonin showed no relationship with depression — and found that lowered serotonin concentration was associated with antidepressant use (n = 1,869).
• A meta-analysis of tryptophan depletion studies found no effect in most healthy volunteers (n = 566), with weak evidence of an effect in those with a family history (n = 75).
• The two largest and highest-quality studies of the SERT gene — a genetic association study of 115,257 people and a collaborative meta-analysis of 43,165 — revealed no evidence of an association with depression, and none of an interaction between genotype, stress and depression.

That last line matters more than it looks. The gene-environment interaction it refers to was, for years, one of the most cited findings in psychiatry. In samples large enough to settle it, it did not hold.

The rebuttal, which belongs on the same page

This review was contested, seriously and in print, and a page that showed you only one side of that would be doing exactly what it criticises.

In 2023, Jauhar and colleagues published a direct response in the same journal — same volume, a few pages earlier. Verbatim [2]:

"We present reasons for why this conclusion is overstated, including methodological weaknesses in the review process, selective reporting of data, over-simplification, and errors in the interpretation of neuropsychopharmacological findings."

Their argument focuses on two of the six areas — tryptophan depletion and serotonergic molecular imaging — as the ones most relevant to investigating serotonin, and contends the umbrella review handled them poorly [2].

And here is our own paper on the paper. In the Resolv library, Moncrieff 2022 scores 72 out of 100 — strong. It earns that for being unfunded, pre-registered, and published in a top-quartile journal. It loses points on allegiance, because the lead author is the most prominent proponent of the position the review supports. That is not an accusation; it is a recorded fact about the paper, and our rubric applies it to everyone. If the score bent for conclusions we found congenial it would be worthless. We wrote up how that scoring works at how we score evidence.

So: a strong, contested review. Not a settled matter, and not a fringe one either.

What patients were told, and where that came from

The scientific question is one thing. The public sentence is another, and it has its own history.

In 2005, Lacasse and Leo published an analysis in PLoS Medicine of the gap between SSRI advertising and the scientific literature. The journal's own summary of it is one line long, and it is the whole story [3]:

"Many ads for SSRI antidepressants claim that the drugs boost brain serotonin levels. Lacasse and Leo argue there is little scientific evidence to support this claim."

That was published in 2005 — seventeen years before the umbrella review. The distance between what the literature supported and what patients were being told was noticed, in a major open-access journal, while the advertising was still running.

We are going to be careful about the word for this. It is not "lie." That word claims knowledge of intent, and intent is exactly the thing we cannot document from the outside. It also invites the easiest possible rebuttal, and would put every sourced claim on this page at risk to buy one satisfying adjective.

The accurate description is a story that outlived its evidence. A hypothesis became a shorthand. The shorthand was clinically useful — it moved depression from character defect to medical condition, which was a genuine gain for a lot of people. It was reinforced by advertising that had every commercial reason to keep it simple. And it stayed in circulation, in doctors' offices and pamphlets and television, long after the research beneath it had stopped supporting it.

Nobody had to lie for that to happen. Simplifications are sticky, especially the kind and useful ones.

Why none of this tells you your medication doesn't work

This is the section to read twice.

Whether a drug works and why it works are separate questions, answered by separate evidence. A mechanism story can be wrong while the treatment is effective. Aspirin was in use for roughly seventy years before anyone worked out prostaglandins. Lithium has been treating bipolar disorder since 1949 and its mechanism is still not settled. Getting the "why" wrong is embarrassing for the textbook; it is not evidence about the outcome.

And on the outcome, the evidence is not thin. The largest antidepressant comparison ever conducted — a network meta-analysis of 21 antidepressant drugs, published in The Lancet in 2018 — found antidepressants more effective than placebo for adults with major depressive disorder, and ranked them on effectiveness and tolerability [4]. We publish that study's caveats too: it was publicly funded, and some co-authors disclosed lecture or consultancy fees from manufacturers.

So the honest summary of both literatures together is:

• The serotonin-deficiency explanation does not have the support it was given in public. (Contested, but seriously argued.)
• Antidepressants outperform placebo on average in the largest trials we have. (Also contested at the margins, and much better supported.)

Both of those can be true at once, because they are answers to different questions. Anyone using the first to argue the second is doing something the evidence does not license.

What it does change

Not the pharmacology. Two other things.

Informed consent. If you agreed to a medication because you were told it corrects a deficiency, you agreed on a basis that the research does not clearly support. You might well make the same decision knowing what the evidence actually says — plenty of people would, and reasonably. But it should be your decision, made on the real footing.

Self-understanding. "You have a chemical imbalance" is a story about a broken part. When the treatment then only partly works, that story quietly implies a broken person. A more accurate framing — we have treatments with measurable average effects and an incompletely understood mechanism — is less tidy and considerably kinder about what a partial response means.

What this is emphatically not

• Not a reason to stop. A randomised trial in UK primary care found 56% of people randomised to discontinue relapsed within a year, against 39% of those who continued [5]. That is a reason to stop deliberately, with a plan, if you stop at all — not a reason never to stop, and not a reason to stop today. The full stopping evidence, including withdrawal as distinct from relapse, is at stopping antidepressants: withdrawal and relapse.
• Not proof that psychiatry is a fraud. It is proof that a specific popular explanation ran ahead of its evidence, which happens across all of medicine and is normally corrected in journals nobody reads.
• Not settled science in the other direction either. The umbrella review has a published rebuttal by a large group of researchers. We have shown you both, and neither of us gets to skip that step.

What to ask your prescriber

• When you say this medication helps, what outcome are we expecting to see, and by when?
• What is the evidence that it works for my situation, separate from the explanation of how?
• If it only partly helps, what does that tell us — and what is the next option?
• What is the plan for reviewing whether I still need it?
• If I ever want to stop, what taper and check-ins would you want in place first?

Sources

1. Moncrieff J, Cooper RE, Stockmann T, Amendola S, Hengartner MP, Horowitz MA. The serotonin theory of depression: a systematic umbrella review of the evidence. Molecular Psychiatry, published online 2022; 2023;28(8):3243–3256. PMID 35854107. PROSPERO CRD42020207203. doi:10.1038/s41380-022-01661-0 — accessed 2026-08-25.
2. Jauhar S, Arnone D, Baldwin DS, Bloomfield M, Browning M, Cleare AJ, Corlett P, Deakin JFW, et al. A leaky umbrella has little value: evidence clearly indicates the serotonin system is implicated in depression. Molecular Psychiatry, 2023;28(8):3149–3152. PMID 37322065. doi:10.1038/s41380-023-02095-y — accessed 2026-08-25.
3. Lacasse JR, Leo J. Serotonin and Depression: A Disconnect between the Advertisements and the Scientific Literature. PLoS Medicine, 2005;2(12):e392. PMID 16268734. Open access. doi:10.1371/journal.pmed.0020392 — accessed 2026-08-25.
4. Cipriani A, Furukawa TA, Salanti G, 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. The Lancet, 2018. doi:10.1016/S0140-6736(17)32802-732802-7). Our summary: /r/51673.
5. Lewis G, Marston L, Duffy L, et al. Maintenance or Discontinuation of Antidepressants in Primary Care (ANTLER). New England Journal of Medicine, 2021. PMID 34587384. doi:10.1056/NEJMoa2106356. Our summary: /r/51676.

Related reading

• Sertraline (Zoloft): what the evidence actually shows
• Where diagnoses come from
• What FDA approval actually means — and what it doesn't
• How we score evidence