Psilocybin for depression: evidence and limitations
Several randomized trials report short-term improvements in depression symptoms with supervised psilocybin and psychological support. Results differ across populations and comparators, and some trials have not met their primary endpoint. Longer-term benefit, safety and performance against established treatments remain important research questions.
This is a selected research guide, not a systematic review or individualized treatment advice. No independent clinical review is claimed. Study interventions are supervised research protocols, not evidence about unsupervised mushroom use.
Depression studies compared by design and outcome
Major depressive disorder (MDD) and treatment-resistant depression (TRD) are different study populations. Read each result with its comparator, endpoint and follow-up. A registration describes a study; a publication reports findings.
| Report and population | Design and sample | Endpoint and finding | Interpretation limit |
|---|---|---|---|
| Mertens et al. · EPISODE · 2026 Treatment-resistant depression; two German centers PubMed 41848690Tracker record and citations NCT04670081 |
Phase 2b randomized trial; 25 mg, 5 mg and nicotinamide comparisons with psychotherapy 144 randomized; 142 in the primary efficacy analysis |
Primary endpoint not met HAMD17 response at week 6, before the second administration Primary comparison not significant: response 17.0% with 25 mg versus 10.6% with nicotinamide; adjusted odds ratio 1.73 (95% CI 0.53–6.23), p=0.19. |
The first hierarchical test failed; secondary symptom changes are exploratory. Two serious adverse reactions followed 25 mg, including one case of hallucinogen persisting perception disorder. |
| Rucker et al. · NHS feasibility trial · 2026 Treatment-resistant major depression; one NHS site in England PubMed 42562964Tracker record and citations |
Randomized, double-blind feasibility trial; psilocybin versus placebo with psychological support 60 randomized; 59 completed MADRS at all follow-up visits |
Feasibility study Primary outcomes: recruitment, retention and estimation of MADRS variance Week-3 adjusted MADRS difference −10.41 points (95% CI −14.86 to −5.95), favoring psilocybin; the report describes persistence at week 6. |
An encouraging symptom estimate from a feasibility study is not a confirmatory efficacy endpoint. A single site and six-week follow-up limit generalization. |
| Yngwe et al. · Stockholm trial · 2026 Moderate to severe recurrent major depressive disorder PubMed 42138922Tracker record and citations NCT04630964 |
Randomized, double-blind trial; psilocybin versus niacin with five support sessions 35 randomized: 17 psilocybin, 18 niacin |
Primary endpoint met Change in clinician-rated MADRS at day 8; follow-up through day 365 Day-8 difference −7.27 points (95% CI −12.89 to −1.65), p=0.01. The clinician-rated difference was not significant at day 365: −3.68 (95% CI −9.30 to 1.94), p=0.20. |
A small sample; later self-reported outcomes are secondary and differ from the clinician-rated endpoint. Two psilocybin participants reported persistent severe anxiety requiring medical attention. |
| Raison et al. · 2023 Major depressive disorder; 11 US research sites PubMed 37651119Tracker record and citations NCT03866174 |
Phase 2 randomized trial; psilocybin versus niacin with psychological support 104 randomized: 51 psilocybin, 53 niacin |
Primary endpoint met Change in centrally rated MADRS from baseline to day 43 Between-group difference −12.3 points (95% CI −17.5 to −7.2), p<0.001, favoring psilocybin. |
Six-week follow-up and selected eligibility. Participants using psychotropic medication could enter following taper. No serious treatment-emergent adverse events were reported, but overall and severe adverse events were more frequent with psilocybin. |
| von Rotz et al. · 2023 Major depressive disorder; Zürich, Switzerland PubMed 36636296Tracker record and citations NCT03715127 |
Randomized, double-blind trial; psilocybin versus placebo with psychological support 52 participants; 26 per group |
Primary outcomes favored psilocybin Change in MADRS and Beck Depression Inventory at day 14 Greater symptom reduction with psilocybin; reported MADRS between-group effect size d=0.97, p=0.0011. |
The primary assessment was only two weeks after treatment. The selected sample and single-center design do not establish long-term effectiveness in routine care. |
| Goodwin et al. · COMP001 · 2022 Treatment-resistant depression PubMed 36322843Tracker record and citations NCT03775200 |
Phase 2 randomized, double-blind trial; 25 mg and 10 mg compared with 1 mg, with psychological support 233: 79 in the 25 mg group, 75 in the 10 mg group, 79 in the 1 mg group |
25 mg primary comparison significant Change in MADRS at week 3; sustained response assessed through week 12 25 mg versus 1 mg: −6.6 MADRS points (95% CI −10.2 to −2.9), p<0.001. The 10 mg comparison was not significant. |
Sustained response at week 12 did not support the short-term result. Adverse events occurred in 77%; suicidal ideation, behavior or self-injury occurred across dose groups. This does not establish their cause. |
| Carhart-Harris et al. · 2021 Long-standing, moderate to severe major depressive disorder PubMed 33852780Tracker record and citations NCT03429075 |
Phase 2 randomized trial; psilocybin treatment versus daily escitalopram, with psychological support in both groups 59: 30 assigned psilocybin, 29 escitalopram |
Primary endpoint not met Change in self-rated QIDS-SR-16 at week 6 Primary between-group difference −2.0 points (95% CI −5.0 to 0.9), p=0.17; no statistically significant difference. |
Several secondary outcomes favored psilocybin without correction for multiple comparisons. The primary result establishes neither superiority nor equivalence to escitalopram. |
| Davis et al. · 2021 Major depressive disorder; participants not taking antidepressants PubMed 33146667Tracker record and citations NCT03181529 |
Randomized immediate-treatment versus waiting-list trial with supportive psychotherapy 27 randomized; 24 completed treatment and post-session assessments |
Positive waiting-list comparison GRID-HAMD depression ratings at one and four weeks after the intervention Depression ratings were lower in the immediate-treatment group than in the waiting-list group at the corresponding assessments. |
A small evaluable sample and a waiting-list comparator. The design does not separate drug effects from expectancy, therapist contact and the treatment setting. |
Export 8 indexed references as BibTeXRIS for citation managers
Each source link remains available even if its record has not yet been indexed or has been withheld for curation. The exports contain only the visible indexed references, not every study on this page.
How to read a positive or inconclusive result
Start with the primary endpoint
A trial specifies its main test in advance. In EPISODE, the primary response comparison was not significant; secondary symptom-score changes remain exploratory. The escitalopram comparison also did not meet its primary test. Positive secondary numbers should not replace these results. EPISODE · Escitalopram trial.
Compare like with like
MADRS, HAMD and QIDS are different depression scales. A symptom-score difference, a response percentage and remission are different outcomes. Even studies using MADRS can differ in population, comparator and timing. These figures should not be ranked as though they measure the same treatment effect.
Keep the treatment setting in view
The selected studies included psychological support or psychotherapy. A waiting-list comparison cannot isolate the pharmacological contribution from the wider intervention. Noticeable psychedelic effects also make masking a challenge; a nominally blinded design alone does not establish successful masking.
Separate short-term change from durability
A significant result after days or weeks cannot establish a sustained benefit after a year. In the Stockholm trial, the clinician-rated difference was significant at day 8 but not day 365. The later self-report pattern involved different, secondary assessments. Stockholm report.
Phase 3 results and sponsor announcements
Compass Pathways reported that COMP006 met its six-week primary endpoint, with a −3.8-point MADRS difference for its 25 mg versus 1 mg regimen. This statement appears in the company's 17 February 2026 SEC filing. It is a sponsor-reported topline result and is not treated here as a substitute for a full peer-reviewed report.
The sponsor also released COMP005 52-week data on 9 September 2026. Part C was open-label and allowed additional treatment. Its baseline changes should not be interpreted as a blinded one-year comparison or as proof that one administration produces a year of benefit.
Check the original reports for methods, missing data, adverse events and analysis populations. Development milestones and efficacy evidence are separate questions; an announcement does not establish regulatory approval.
Safety findings and who the evidence describes
Trials select participants and provide monitoring and follow-up. Their results do not automatically apply to people excluded by a protocol. The Raison trial excluded, among other groups, people with a history of psychosis or mania and active suicidal intent; medication users could enter following taper. Read the study context.
Small studies cannot rule out uncommon harms. The Goodwin trial recorded adverse events and suicidal ideation, behavior or self-injury across groups; EPISODE reported serious adverse reactions; the Stockholm study described severe persistent anxiety in two participants. These reports need their full clinical context and should neither be ignored nor converted into a causal risk estimate this guide cannot support.
Do not change psychiatric medication to qualify for a study based on this website. Eligibility and any medication requirements must be discussed with the study team and the prescriber responsible for your care. A registry listing does not determine personal eligibility.
Read how to investigate participation or explore the depression registry collection, which includes recruiting and non-recruiting studies.
Questions about psilocybin depression research
Does psilocybin work for depression?
Several selected controlled studies found short-term symptom improvement with supported psilocybin interventions. Others did not meet their primary endpoint. The evidence supports further investigation, with conclusions tied to the population, comparator and time point studied. It does not predict an individual's outcome.
Has it been shown to be better than antidepressants?
The 2021 escitalopram trial did not demonstrate a significant difference on its primary outcome. Its secondary results do not establish general superiority, and a nonsignificant result is not a demonstration of equivalence. Compare the primary result.
Are treatment-resistant depression results interchangeable with MDD results?
No. TRD studies enroll people with inadequate response to prior treatment according to each protocol. Their baseline severity and prior treatment differ from other MDD samples. Keep these populations separate when screening studies or interpreting effect sizes.
Where can I find ongoing depression studies?
Our depression trial collection groups registry records by condition. For recruitment discovery use the exact-status recruiting collection and confirm contacts, eligibility and local availability at ClinicalTrials.gov.
How this guide was assembled
The selection covers eight primary randomized reports in MDD or TRD, chosen to illustrate waiting-list, placebo, active-treatment and feasibility designs, alongside relevant 2026 findings. It is a purposive educational selection, not a complete search, risk-of-bias assessment, systematic review or meta-analysis. Cancer-related distress, bipolar depression, observational programs and secondary analyses are outside this comparison.
Bibliographic identity, sample descriptions, endpoints and numerical findings were checked against primary-report records on 7 October 2026. Source records and sponsor disclosures are linked separately. The annotations are original summaries of factual findings; unverified abstracts and source text are not redistributed. No formal certainty grade, pooled estimate or independent clinical-review credit is assigned.
The eight report DOIs and their linked PubMed records define this guide's source set. Database updates can add a tracker link but do not silently update the interpretation or the source-check date. Report a correction with its authoritative source.