Skip to content
Psilocybin Research

The best psilocybin trial is about drinking, not depression

It is double-blind, it used an active placebo, and it met its primary endpoint. It also gets a small fraction of the coverage the depression trials get. That ordering is almost exactly backwards.

The strongest evidence in the fieldReviewed

The short answer

The best-designed psilocybin trial published to date is in alcohol use disorder, and almost nobody knows it exists. The 2022 trial by Bogenschutz and colleagues randomized 95 people, was double-blind, and compared psilocybin against diphenhydramine — an active placebo that produces noticeable effects of its own, rather than an inert pill. Heavy drinking days were 9.7% against 23.6% over 32 weeks. That design addresses the weakness that undermines most psychedelic research, which is why this single trial carries more weight than the larger depression literature. It is still one trial of 95 people, both arms received twelve psychotherapy sessions, and psilocybin is not approved for anything.

Participants randomized
95
Comparator
Active placebo
Heavy drinking days
Abstinence
47.9% vs 24.4%
Effect size
Hedges g=0.52

The ledger

The trial, with what qualifies it

One row, because there is one trial. The absence of a second is itself the most important thing on this page.

The 2022 randomized trial of psilocybin for alcohol use disorder, with design, sample size, comparator and primary endpoint
TrialDesignParticipantsPrimary endpointResult and limitation
Psilocybin for alcohol use disorder2022 · Alcohol use disorderSource ↗

Randomized, double-blind trial with 12 psychotherapy sessions in both arms

ControlledDouble-blind

Compared against: Active placebo (diphenhydramine)

95Met

Heavy drinking days 9.7% versus 23.6% over 32 weeks — a mean difference of 13.9 percentage points (95% CI 3.0–24.7, p=0.01, Hedges g=0.52). Abstinence 47.9% versus 24.4%.

Read it with this: The blind was still not fully maintained, and the psilocybin effect cannot be separated from the twelve psychotherapy sessions both arms received.

Every other trial we track, including the depression trials and the conditions with no published data at all, is on the research hub.

Why the active placebo matters more than the result

Psychedelic research has one methodological problem that sits underneath all the others. A person who takes psilocybin knows they have taken psilocybin. The effects are unmistakable, they last hours, and no amount of procedural care hides them. When the comparison arm receives an inert pill, everyone in the trial can work out their assignment within an hour of dosing, and the ratings that follow are shaped by expectation as much as by pharmacology.

This trial used diphenhydramine as the comparator.[1] Diphenhydramine produces noticeable effects — sedation, dry mouth, a clear sense that something has been taken. A participant who felt something could not conclude from that alone that they had received psilocybin. The inference that breaks the blind everywhere else in this field was, at minimum, made harder.

That is the reason to rank this trial above the depression literature, and it is worth being explicit about the comparison. The Compass Phase 3 trials used a placebo and a 1 mg dose as comparators.[2] The most-quoted trial in the field compared psilocybin against escitalopram and did not meet its primary endpoint.[3] Attention in this field has tracked condition and press release rather than design. On design, the drinking trial wins.

The numbers

Ninety-five adults with alcohol use disorder were randomized. Over the 32 weeks following the first dose, the percentage of heavy drinking days was 9.7% in the psilocybin group and 23.6% in the placebo group — a mean difference of 13.9 percentage points, 95% confidence interval 3.0 to 24.7, p=0.01, Hedges g=0.52. Complete abstinence was reported by 47.9% of the psilocybin group against 24.4% of the placebo group.[1]

Read the confidence interval rather than the point estimate. It runs from 3.0 to 24.7 percentage points, which is consistent with a difference large enough to change a life and with one small enough to be marginal. A trial of 95 people cannot narrow that range further. Hedges g of 0.52 is a moderate effect size by the usual conventions, and a moderate effect in a condition where treatment often fails is a serious finding.

What the trial cannot show

These belong in the same paragraph as the result, not in a limitations section nobody reads.

The blind was still not fully maintained. An active placebo makes the inference harder; it does not make it impossible. The authors report that participants and staff could still often tell, and the expectancy problem is reduced rather than solved.

The psilocybin effect cannot be isolated. Both arms received twelve psychotherapy sessions. What the trial establishes is that psilocybin added to that psychotherapy outperformed an active placebo added to the same psychotherapy. Anyone describing this as evidence for psilocybin alone is describing a trial that has not been run.

It is one trial, not a literature. Ninety-five people at a small number of sites, with no independent replication. Single trials of this size have failed to replicate across every area of medicine, and there is no reason to assume this one is exempt. A second trial with an active placebo, run by a different group, would do more for the case than any amount of further discussion of this one.

Psilocybin is not approved for anything. The FDA has approved no psilocybin medicine, for alcohol use disorder or any other condition, and it remains a Schedule I controlled substance federally. This trial describes a research intervention that is not available as treatment.

This is not a replacement for existing treatment

Alcohol use disorder already has treatments with a far larger evidence base than one trial of 95 people. Approved medicines exist, behavioral treatments exist, and unlike psilocybin they can be prescribed, are generally covered by insurance, and can be started this week. None of that is displaced by a promising trial of something that is not available.

The failure mode we see most often is a person reading a headline about this trial and treating it as a reason to wait — to postpone starting treatment until psilocybin becomes available, on the assumption that it will be better. There is no evidence supporting that comparison. No trial has compared psilocybin against an approved treatment for alcohol use disorder, so the question of which works better has not been asked, let alone answered.

Common questions

Questions about psilocybin and drinking

Does psilocybin work for alcohol use disorder?
One well-designed randomized trial says it may. In the 2022 trial by Bogenschutz and colleagues, 95 people with alcohol use disorder received either psilocybin or diphenhydramine as an active placebo, alongside twelve psychotherapy sessions in both arms. Heavy drinking days were 9.7% in the psilocybin group against 23.6% in the placebo group over 32 weeks — a mean difference of 13.9 percentage points, 95% CI 3.0 to 24.7, p=0.01, Hedges g=0.52. That is one trial of 95 people. It is a strong signal, not a settled question.
Why does this trial deserve more weight than the depression trials?
Because of the comparator. Psychedelic trials usually fail at blinding: participants can tell they received a psychedelic because the effects are unmistakable, and expectation then inflates the apparent benefit. This trial used diphenhydramine, which produces noticeable effects of its own, so participants could not simply infer their assignment from the fact that something happened. That addresses the field's central methodological weakness, which the depression trials mostly do not.
Was it the psilocybin or the psychotherapy?
The trial cannot separate them, and the authors say so. Both arms received twelve psychotherapy sessions, so what the result establishes is that psilocybin added to that psychotherapy outperformed diphenhydramine added to the same psychotherapy. It does not establish what psilocybin does on its own, and nobody should present the drug as the whole of the intervention.
Is psilocybin an approved treatment for alcoholism?
No. Psilocybin is not approved by the FDA for alcohol use disorder or for any other condition, and it remains a Schedule I controlled substance under federal law. Nothing described on this page is available as a prescribed treatment anywhere in the United States. Oregon and Colorado operate supervised-access programs, but those are not medical treatment and are not what the trial studied.
Should I stop my current treatment to try psilocybin?
No. Established treatments for alcohol use disorder exist, are approved medicines, and are generally covered by insurance — which psilocybin is not. Alcohol withdrawal can also be medically dangerous and in some cases life-threatening, and needs supervision rather than improvisation. A single trial of 95 people is not a reason to change or stop anything that is currently working.

References

  1. 1.Bogenschutz MP, Ross S, Bhatt S, et al. Percentage of Heavy Drinking Days Following Psilocybin-Assisted Psychotherapy vs Placebo in Adult Patients With Alcohol Use Disorder: A Randomized Clinical Trial. JAMA Psychiatry. 2022;79(10):953–62.The best-designed trial in the field: n=95, double-blind, with an ACTIVE placebo (diphenhydramine) rather than an inert one. Heavy drinking days 9.7% vs 23.6% over 32 weeks (mean difference 13.9%, 95% CI 3.0–24.7, p=0.01, Hedges g=0.52). Abstinence 47.9% vs 24.4%. The authors note the blind was still not fully maintained, and psilocybin's effect cannot be separated from the 12 accompanying psychotherapy sessions.
  2. 2.Compass Pathways. COMP005 and COMP006 Phase 3 topline results. 2025–2026.Both trials met their primary endpoint. The differences are modest: −3.6 MADRS points vs placebo (COMP005) and −3.8 for 25 mg vs 1 mg (COMP006).
  3. 3.Carhart-Harris R, et al. Trial of Psilocybin versus Escitalopram for Depression. N Engl J Med. 2021;384:1402–11.The primary outcome was NOT statistically significant (difference −2.0, 95% CI −5.0 to 0.9, p=0.17). The widely quoted response and remission figures are unadjusted secondary outcomes; the authors state no clinical conclusions can be drawn from them.