Psilocybin for depression: what the clinical trials found

Psilocybin for depression has been tested in a handful of randomized clinical trials, and the results are not uniform. A waiting-list trial in 2021 reported large score drops (Davis et al., 2021). A head-to-head trial against the antidepressant escitalopram did not find a significant difference on its primary outcome (Carhart-Harris et al., 2021). A 2026 German trial, called by its own authors an inconclusive trial, missed its primary endpoint (Mertens et al., 2026).

This page describes what each study did and found, and where its authors and critics see weak points. It is not medical advice, it does not say what anyone should do, and it covers supervised research only. Results from clinics, from retreats or from people acting alone are a different matter and were not tested in these studies.

Browse all studies: our psilocybin studies library lists the major published psilocybin studies, each with a plain-English summary of what it found.

What did the main psilocybin depression trials find?

The table below puts the five trials side by side. Scores come from different scales, so the numbers are not directly comparable across rows. Every trial gave psilocybin with psychological support in a clinical setting. We leave out dose amounts on purpose, because this page is about what was found, not how to use anything.

Three things stand out. The comparators differ widely: a waiting list, a daily antidepressant, a very low dose and an inactive-looking pill. The sample sizes are small to moderate. And the earliest, smallest trial reported the largest effect sizes, while the most recent one missed its primary endpoint.

How was each trial designed, and what did it find?

Davis et al., 2021: the waiting-list trial

This Johns Hopkins study randomized 27 adults with major depressive disorder to treatment right away or after an 8-week delay. Adults were aged 21 to 75, not taking antidepressant medication, and had no history of psychotic disorder, serious suicide attempt or hospitalization. 24 completed the intervention and the week 1 and week 4 assessments. The primary outcome was a clinician-rated depression scale, and the reported effect sizes were large (Cohen d of 2.5 and 2.6) (Davis et al., 2021).

The authors state the limits plainly. The sample was small and predominantly White non-Hispanic, follow-up was short, and the delayed-treatment group did not control for preparation, rapport, integration meetings or expectancy. They also note that placebo and active-control designs have their own limits because psilocybin’s effects are so noticeable. They called for larger and more diverse samples, longer follow-up and a placebo control. The paper reports no serious adverse events, and about a third of sessions included a mild to moderate transient headache (Davis et al., 2021).

Carhart-Harris et al., 2021: psilocybin versus escitalopram

This was the first trial in this set to compare psilocybin directly with an established antidepressant. It enrolled 59 people with long-standing moderate-to-severe major depressive disorder, split 1:1. One group received two psilocybin sessions three weeks apart and six weeks of daily placebo. The other received two sessions of a very low psilocybin dose and six weeks of daily escitalopram. All received psychological support (Carhart-Harris et al., 2021).

The primary outcome, change in a 16-item self-rated scale at week 6, was -8.0 with psilocybin and -6.0 with escitalopram, a difference that was not statistically significant (95% CI -5.0 to 0.9). Response rates were 70% and 48%; remission rates were 57% and 28%. The authors wrote that other secondary outcomes generally favoured psilocybin but were not corrected for multiple comparisons, and that the trial did not show a significant difference on the primary measure. Adverse events were similar between groups (Carhart-Harris et al., 2021).

Goodwin et al., 2022: the COMP360 phase 2 trial

Goodwin and colleagues randomized 233 adults with treatment-resistant depression to a single session of one of three doses of a proprietary synthetic formulation of psilocybin, with psychological support. The lowest dose served as the control. The company COMPASS Pathfinder funded the study. On the primary outcome (change in a clinician-rated scale at week 3), the higher dose beat the control by 6.6 points (95% CI -10.2 to -2.9). The middle dose did not differ significantly from the control (Goodwin et al., 2022).

The abstract reports that response and remission at 3 weeks supported the main result, but sustained response at 12 weeks did not. Adverse events occurred in 77% of participants (179 of 233) and included headache, nausea and dizziness. Suicidal ideation, suicidal behavior or self-injury occurred in all dose groups. The authors concluded that larger and longer trials, including comparison with existing treatments, are needed (Goodwin et al., 2022). For the regulatory side of this formulation, see our explainer on COMP360 and the FDA.

Raison et al., 2023: one session versus niacin

This phase 2 trial ran at 11 US sites. 104 adults with major depressive disorder of at least 60 days’ duration and moderate or greater severity were randomized 1:1 to psilocybin or niacin, used as an active placebo. Both came in identical-looking capsules and both were given with psychological support. People with psychosis, mania, active substance use disorder or active suicidal ideation with intent were excluded (Raison et al., 2023).

The primary outcome was change in a clinician-rated depression score from baseline to day 43. Psilocybin was associated with a 12.3-point larger drop than niacin (95% CI -17.5 to -7.2). More participants given psilocybin had sustained response, but not remission. No serious treatment-emergent adverse events occurred, but 44 of 50 psilocybin participants (88%) and 33 of 54 niacin participants (61%) reported at least one adverse event, and the paper reports a higher rate of severe adverse events with psilocybin (Raison et al., 2023).

The authors list limits. The success of blinding was not assessed, and psilocybin’s acute effects likely produced some functional unblinding. Niacin may have made the comparator more believable and raised the placebo response. The sample was predominantly White, non-Hispanic and from upper socioeconomic groups (Raison et al., 2023).

Mertens et al., 2026: the EPISODE trial

EPISODE is a German phase 2b trial published in JAMA Psychiatry in 2026. It was triple-blinded (participant, investigator and rater) and used an active placebo. Adults aged 25 to 65 with treatment-resistant depression, withdrawn from antidepressant medication, were randomized to four sequences across two centres. We compare the three arms with analysed data at week 6: higher-dose psilocybin (47), low-dose psilocybin (48) and nicotinamide (47) (Mertens et al., 2026).

The primary endpoint, response at week 6, was 17.0%, 12.5% and 10.6%. The first planned comparison (higher dose against nicotinamide) was not significant (adjusted odds ratio 1.73, 95% CI 0.53 to 6.23, P = .19), so formal testing stopped. The key secondary results gave exploratory evidence of a clinically meaningful effect with the higher dose. The authors call the trial inconclusive (Mertens et al., 2026).

They also report safety signals. Suicidal ideation on dosing days was reported by 4% after the higher dose versus 1% to 2% in comparator conditions. Two serious adverse reactions followed the higher dose, including one case of hallucinogen persisting perception disorder, a lasting disturbance in perception (Mertens et al., 2026).

Its limitations section names functional unblinding (86% correctly identified the higher dose), a small centre effect, an overestimated effect in the power calculation, a socioeconomically and ethnically homogeneous sample, no follow-up beyond 12 weeks, and exclusion of people at high suicide risk (Mertens et al., 2026). A separate 6- and 12-month follow-up reported lasting average score changes, but it was naturalistic: after week 12, participants could seek other treatments, and no controlled comparison was kept (Mertens et al., 2026, follow-up).

Can participants tell whether they received psilocybin?

This is the central methodological question in this field. Psilocybin produces strong, noticeable effects, so participants and staff can often guess who got what. That is called functional unblinding. If people expect a drug to help, believing they received it may itself change self-reported and even clinician-rated scores, a problem called expectancy.

A 2021 review in Expert Review of Clinical Pharmacology looked at this directly. It reviewed the literature on expectancy and blinding and ran systematic searches (Medline, PsycInfo and Embase, January 1990 to November 2020) for psychedelic trials. The authors found that reported psychedelic trials had generally not reported pre-trial expectancy, nor whether blinding had worked. They argued that effect sizes in psychedelic trials are likely overestimated and urged caution when reading them (Muthukumaraswamy et al., 2021).

The trials above bear this out in different ways. Raison’s authors say blinding success was not assessed and unblinding was likely. EPISODE did measure it, and most participants identified the higher dose. In EPISODE’s own analysis, the authors suggested unblinding of the first dose played a negligible role in week-6 depression outcomes, while also stating that expectancy effects could not be determined. Davis’s waiting-list design did not blind participants to their treatment, though clinician raters were blinded.

A 2025 meta-analysis in JAMA Network Open examined this from another angle. It pooled 4 psilocybin trials (373 participants) and compared their control arms with control arms in 2 esketamine and 11 SSRI trials. Control groups in psilocybin trials improved less (standardized mean change 0.50) than control groups in SSRI (1.00) or esketamine (1.12) trials, while the active groups improved by similar amounts (1.21, 1.28 and 1.43). The authors say this might indicate psilocybin’s efficacy is overestimated compared with that of SSRIs and esketamine. That is an indirect, cross-trial comparison, not a head-to-head test (Hieronymus et al., 2025).

What do pooled analyses say?

A 2024 BMJ systematic review and meta-analysis by Metaxa and Clarke pooled randomized trials of psilocybin for depression. It rated all included studies at moderate risk of bias, and said further research is needed on the influence of expectancy effects, moderating factors and treatment delivery (Metaxa and Clarke, 2024).

That paper has since been corrected. After publication, concerns were raised about the calculation of standardised mean differences and about data from the same study being counted more than once. The authors agreed there were errors, corrected the estimates and removed duplicated results. The correction says the main messages are unaltered and describes the research as preliminary, with no clinical implications (BMJ correction, 2025). We do not quote the pooled effect size here, because the estimates were corrected and we could not verify the corrected figures.

What do these trials not show?

• They do not show that psilocybin is an approved treatment. In Canada it is not (see the next section).
• They do not test self-treatment. Participants were screened, supervised during sessions, given psychological support and followed up by clinicians, and each trial excluded some groups of people. Findings from those settings do not transfer to use outside them.
• They do not show long-term outcomes. Primary outcomes were measured within about six weeks of the first session, and no controlled comparison in these trials ran past week 12. In Goodwin’s trial, for example, the primary end point was week 3 and the double-blind follow-up for sustained response ran to week 12 (Goodwin et al., 2022). Uncontrolled follow-ups exist, but they lack a comparison group (Mertens et al., 2026, follow-up).
• They do not show who would be harmed. Several trials excluded people with a history of psychosis, mania, serious suicide attempts or high suicide risk, so these studies say little about those groups.
• They do not establish how large the effect is. Blinding and expectancy may inflate it, and the estimates range from non-significant (EPISODE, escitalopram comparison) to large (waiting-list design).
• They do not reflect the full population. Several samples were predominantly White and, where stated, from higher-income groups.
• They do not separate the drug from the therapy. Every trial paired psilocybin with psychological support, and most compared against a control that also received support.

What adverse events did the trials report?

Adverse events were common and were reported in every trial. Headache, nausea and dizziness appeared in Goodwin’s trial, where 77% of participants reported adverse events (Goodwin et al., 2022). Raison’s trial reported 88% with psilocybin and 61% with niacin, and a higher rate of severe events with psilocybin, though none serious (Raison et al., 2023). Davis reported transient headaches and difficult emotional experiences, and a short rise in blood pressure in one session (Davis et al., 2021). Suicidal ideation, suicidal behavior or self-injury was reported across dose groups in Goodwin’s trial, and EPISODE reported higher suicidal ideation on dosing days after the higher dose and one case of hallucinogen persisting perception disorder (Goodwin et al., 2022; Mertens et al., 2026).

EPISODE’s authors noted that adverse-event rates were higher than in earlier studies, likely reflecting more systematic assessment. They said this highlights the need for standardized risk monitoring (Mertens et al., 2026). Our safety page covers risks in everyday terms.

Is psilocybin an approved treatment for depression in Canada?

As of 24 October 2026, Health Canada’s page (last modified 20 October 2026) states that production, sale and possession of magic mushrooms are illegal in Canada, and that there are no approved therapeutic products containing psilocybin in Canada or elsewhere (Health Canada). Approval positions can change, so check that page for the current position.

The same page notes that Health Canada restored the ability for health care practitioners to request psilocybin through the Special Access Program in January 2022. It describes that program as a route for drugs that have shown promise in clinical trials or are approved elsewhere but not yet authorized for sale in Canada (Health Canada). We have not reviewed how that program works in practice, and this article does not advise on it.

What is still unknown?

• How much of the observed difference comes from the drug, and how much from expectancy, unblinding, preparation and therapy. Authors on both sides say this is unsettled.
• How long any effect lasts when a controlled comparison is kept. The longer follow-ups we found were not controlled.
• How results hold up in larger, more diverse samples, and in groups the trials excluded.
• How psilocybin compares with established treatments. One head-to-head trial found no significant primary difference, and its authors called for larger, longer trials.
• The full safety picture. EPISODE’s authors say standardized risk monitoring is still needed.

Larger, longer trials are what the authors of these studies call for. Until those report, results are best described as early and mixed. For background on the compound itself, see what psilocybin is and what magic mushrooms are. For another condition studied in trials, read our look at psilocybin and alcohol use disorder.

Shroomverse sells mushroom products. This article is general information about clinical research. It is not medical, legal or purchasing advice, and it does not recommend any product.