Psilocybin
& psilocin
Two completed Phase 3 trials, an NDA in preparation, and an effect size worth arguing about.
Psilocybin is the reference compound for the entire modern psychedelic field — the molecule against which every other asset positions itself, including the two in this series. It has the deepest indigenous use history, the cleanest prodrug logic, the largest modern trial programme, and, as of 2026, the only completed pair of Phase 3 trials for a classic psychedelic. It is also the compound whose effect size is most openly debated: roughly −3.6 to −3.8 MADRS points against a low-dose comparator at Week 6. This report treats the prodrug and the active metabolite as distinct pharmacological entities, because the development logic depends on the distinction.
Psilocybin is inactive; psilocin does the work.
Psilocybin is a 4-phosphoryloxy prodrug, dephosphorylated by alkaline phosphatase to psilocin. Nature solved the same stability problem that 4-OH-DiPT's developers solved with a glutarate ester.
It too was a marketed Sandoz medicine.
Hofmann isolated and synthesised it in 1958–59 and Sandoz sold it as Indocybin for psycholytic therapy — a second licensed psychedelic, withdrawn in the same collapse that took Delysid.
Both Phase 3 trials hit — at a modest margin.
COMP005: −3.6 MADRS vs placebo at Week 6. COMP006: −3.8 for two 25 mg doses vs 1 mg (p < 0.001). Statistically robust, clinically arguable — and far smaller than the Phase 2b signal implied.
Two doses beat one, and durability reaches six months.
Response was 25% in COMP005 and 39% in COMP006; responders maintained benefit through at least Week 26. Retreatment was well tolerated — the most useful new operational fact in the programme.
The comparator is a low dose, not a placebo — mostly.
COMP005 used true placebo; COMP006 used 1 mg. Running both designs across a pair of pivotal trials is a deliberate answer to the blinding critique, and a better one than either design alone.
Legal reform has outrun the evidence.
Oregon and Colorado built supervised-use frameworks and dozens of cities decriminalised — all before any Phase 3 read out. Psilocybin is the one compound where state law leads federal science.
From Oaxaca to an NDA
Origins:
the only one that was already in use
Psilocybin differs from every other compound in this series in one fundamental respect: it was not invented. Psilocybe mushrooms had been used ceremonially in Mesoamerica for centuries before any chemist saw them, and that fact carries both scientific weight and an ethical obligation that the field has handled unevenly.
Spanish chroniclers recorded ritual use of teonanácatl — conventionally glossed as “flesh of the gods” — among Nahua peoples in the sixteenth century, and colonial and ecclesiastical authorities suppressed the practice. Use persisted, notably among Mazatec communities in Oaxaca. In 1955 the American banker and amateur mycologist R. Gordon Wasson participated in a velada conducted by the Mazatec curandera María Sabina, and published an account in Life magazine in 1957.1
The consequences for Sabina and Huautla de Jiménez were severe: an influx of outsiders, community ostracism, and by her own account the loss of the practice's integrity. She died in poverty. Every subsequent commercial development of psilocybin — including a Nasdaq-listed NDA — traces its knowledge of the compound's activity to a lineage that received nothing. Benefit-sharing remains an unresolved question in the field, not a settled one.1,2
Wasson sent material to Sandoz, where Albert Hofmann — the same chemist who had made LSD — isolated psilocybin and psilocin in 1958 and achieved total synthesis shortly after.3 Sandoz then did exactly what it had done with LSD: marketed it, as Indocybin, for psycholytic psychotherapy. Psilocybin's shorter duration made it the preferred agent for many European clinicians.
The 1960s American history is largely the Harvard Psilocybin Project. Two studies from it are still cited and both are cautionary. The Good Friday Experiment (Pahnke, 1962) gave psilocybin or placebo to divinity students in a chapel setting and reported markedly more mystical experience in the active group — a genuinely interesting double-blind design, undermined by later disclosure that adverse reactions had gone unreported. The Concord Prison Experiment claimed reduced recidivism; a 1990s re-analysis found the control comparison did not support the claim.2,4 Both are object lessons in how psychedelic research goes wrong: real phenomena, motivated reporting.
Psilocybin was placed in Schedule I in 1970 alongside LSD, and the research stopped for the same reasons and the same length of time.
Mechanism: the prodrug and the drug
Psilocybin — 4-phosphoryloxy-N,N-dimethyltryptamine — is essentially pharmacologically inert at the target. It is rapidly dephosphorylated, principally by alkaline phosphatase, to psilocin (4-HO-DMT), which is the active agent. Every statement about “psilocybin's mechanism” is really a statement about psilocin.3,5
This is worth dwelling on, because it is the same design solution arrived at three separate times. The free 4-hydroxy group of psilocin is chemically labile and subject to first-pass metabolism; masking it as a phosphate ester (psilocybin, natural), an acetate (4-AcO-DMT, semi-synthetic), or a glutarate hemiester (luvesilocin, for 4-OH-DiPT) all solve the same problem. Nature got there first, and the phosphate ester is why psilocybin rather than psilocin is the pharmaceutical.
Target profile
Psilocin is a non-selective serotonin receptor agonist with high affinity at 5-HT2A, and meaningful activity at 5-HT2C, 5-HT1A and 5-HT2B.5,6 As with LSD, the causal role of 5-HT2A in the subjective effects is established in humans by antagonist blockade: ketanserin pre-treatment prevents psilocybin's psychedelic effects.7 Psilocin produces the head-twitch response in rodents, and its potency in that assay is the reference point against which analogues — including 4-OH-DiPT and 4-HO-DPT — are calibrated.6
Two contrasts with 4-OH-DiPT are pharmacologically substantive. Psilocin has appreciable 5-HT2C agonism where 4-OH-DiPT's is reported as much weaker; and psilocin's oral duration of 6–8 hours is roughly double. The N,N-diisopropyl substitution is what buys the brevity, at the cost of a steeper dose–response curve.
Downstream: connectivity and plasticity
Imaging under psilocybin shows reduced default-mode network integrity and increased between-network connectivity, and — in the most methodologically ambitious work — desynchronisation of functional networks with individual-level effects far larger than those produced by a stimulant control.8 Antidepressant response has been associated with post-treatment increases in network flexibility.9
At the cellular level psilocin sits in the psychoplastogen class — promoting dendritic spine formation and synaptogenesis via 5-HT2A, with rapid and persistent structural remodelling shown in cortex.10 The mechanistic account is therefore identical in shape to the one offered for LSD and, by a different receptor route, for methylone: acute receptor engagement, then a plasticity window. Whether the subjective experience is necessary to the outcome or merely concurrent with it is the field's central unresolved question, and psilocybin is where it will be settled.
One problem, three prodrugs
Every 4-hydroxytryptamine in development masks the same labile hydroxyl. The chemistry chosen determines route, onset and — commercially — patentability.
25 mg psilocybin oral is the therapeutic dose across the programme; 1 mg and 10 mg serve as comparators.11,13
6–8 h, onset 20–40 min. A full monitored day, though shorter than LSD.
Session cost sits between LSD and the short-acting tryptamines — the middle of the field's operational spectrum.
The modern revival, 2006–2022
The modern era has a clear starting point. In 2006 Roland Griffiths' group at Johns Hopkins published a rigorous double-blind study showing that psilocybin reliably occasioned mystical-type experiences with sustained positive attributions at 14-month follow-up.11 Its significance was procedural as much as scientific: it demonstrated that this research could be done to a standard, in the US, under Schedule I.
What followed, in rough order of evidentiary strength:
- Cancer-related depression and anxiety (2016). Two simultaneous randomised trials — Griffiths at Hopkins and Ross at NYU — reported large, rapid and sustained reductions in depressed mood and anxiety in patients with life-threatening cancer, with benefit persisting at six months. These remain among the most striking results in the literature and, notably, both used high-dose/low-dose crossover designs.12
- Major depressive disorder. Davis et al. 2021 in JAMA Psychiatry reported large effects in a randomised waiting-list design, and a 2023 randomised placebo-controlled trial in MDD reported significant reductions in depressive symptoms.13 Waiting-list controls inflate effect sizes; the 2023 trial is the better evidence.
- Alcohol use disorder. Bogenschutz et al. 2022 reported reduced heavy drinking days versus active placebo in a randomised trial — the modern successor to the 1960s alcoholism work, and a genuine replication of an old signal.14
- Smoking cessation and OCD. Promising open-label pilots, insufficient controlled evidence. Do not overweight them.15
The pivotal transition came with COMP360, Compass Pathways' synthetic, proprietary psilocybin formulation. Its Phase 2b, published in the New England Journal of Medicine in 2022, randomised 233 patients with treatment-resistant depression to 25 mg, 10 mg or 1 mg with psychological support. The 25 mg arm showed a significant MADRS reduction versus 1 mg at Week 3; the 10 mg arm did not separate.16 That dose-response pattern is what justified taking 25 mg into Phase 3, and the 1 mg comparator became the programme's signature.
One honest observation about the arc: effect sizes have shrunk as designs have tightened. The 2016 cancer trials reported very large effects against low-dose crossover controls in small, highly selected, expectant populations. The Phase 3 differences are a few MADRS points. That is the normal trajectory of a real but moderate treatment effect meeting rigorous methodology — and it is the single most useful thing to know when reading any psychedelic Phase 2 result, including in adjacent programmes.
COMP005 and COMP006
Compass describes the COMP360 Phase 3 programme as the largest randomised, controlled, double-blind psilocybin programme ever conducted.17 Two pivotal trials, both in treatment-resistant depression, both with a Week-6 MADRS primary endpoint, and — importantly — with different comparators.
COMP005 (NCT05624268) is randomised, double-blind and placebo-controlled: 258 participants with TRD across 32 US sites, testing a single 25 mg dose against placebo. It runs in three parts — Part A blinded through Week 6, Part B blinded through Week 26, and a Part C extension.17,18 It was the first Phase 3 study of a synthetic psilocybin and the first classic psychedelic to report Phase 3 efficacy data.17 Result: a −3.6-point MADRS treatment difference at Week 6.19
COMP006 is dose-controlled rather than placebo-controlled, comparing fixed doses. Announced on 17 February 2026: two 25 mg doses administered three weeks apart versus a 1 mg control produced a −3.8-point MADRS difference at Week 6, p < 0.001.20,21
How to read a 3.6-point difference
This is where judgement is required, and where commentary tends to divide along prior commitments. Three points, in tension:
Against: a 3.6–3.8-point MADRS separation is at or near the conventional threshold for clinical meaningfulness, and it is smaller than the Phase 2b and the 2016 academic trials led people to expect. A 25% response rate in COMP005 leaves three quarters of patients without a clinically meaningful response.
For: the population is genuinely refractory — severe depression, many lifetime episodes, chronic current episodes22 — and the comparison in COMP006 is against 1 mg of an active drug in a fully supported therapeutic setting, which is a demanding control. Durability through 26 weeks from one or two doses is a different value proposition from a daily antidepressant, and MADRS at Week 6 does not capture it.
The honest synthesis: the effect is real, reproducible and moderate. Two adequately powered pivotal trials agreeing to within 0.2 MADRS points is a strong result on reproducibility, which is more than most psychiatric assets can claim. Whether moderate is enough will be decided on durability and on the retreatment paradigm rather than on the Week-6 number. COMP360 holds FDA Breakthrough Therapy designation for TRD, with an NDA submission planned for Q4.21
The two pivotal trials, side by side
On the comparator strategy. Running one placebo-controlled and one dose-controlled pivotal trial is the most rigorous answer to the functional-unblinding objection anyone in this field has attempted — stronger than MindMed's two-plus-a-low-dose-arm approach and stronger than a single dose-controlled study. If psilocybin is approved, this design pattern is likely to become the expectation for every psychedelic that follows, including short-acting ones.
Where law overtook evidence
Psilocybin is federally Schedule I in the United States and controlled under the 1971 UN Convention. It is also the only compound in this series where sub-national jurisdictions have built legal access frameworks ahead of any pivotal clinical evidence — a sequence with no real precedent in modern drug regulation.
Oregon passed Measure 109 in 2020, creating a state-regulated supervised psilocybin services programme — not a medical prescription model, but licensed facilitators and service centres, with the first licences issued in 2023. Colorado followed with Proposition 122 in 2022, decriminalising personal use of several natural psychedelics and authorising healing centres. Dozens of US cities, beginning with Denver in 2019, have deprioritised or decriminalised possession.24
The analytical point is not whether these policies are good. It is that they create a two-track reality that a pharmaceutical sponsor has to navigate: an approved COMP360 would enter a market where a legal, cheaper, non-pharmaceutical route to supervised psilocybin already exists in some states. That is a genuinely novel commercial problem, and it does not exist for methylone, LSD or 4-OH-DiPT.
Psilocybin is a natural product described in 1958; the molecule cannot be patented. Compass's protection rests on crystalline polymorph, formulation and method-of-use claims around COMP360, which have been contested by third parties. This is structurally the same position as MM120 (formulation) and luvesilocin (prodrug ester): in psychedelics, nobody owns the molecule, so everyone competes on delivery and process. Judge these programmes on regulatory exclusivity and execution, not patent life.25
On safety and abuse liability, psilocybin's profile closely resembles LSD's: it is not reliably self-administered by animals, produces no withdrawal syndrome, has a wide therapeutic index and no established human lethal dose. The material risks are psychiatric — acute adverse reactions requiring supervision, precipitation of psychosis in vulnerable individuals (hence universal exclusion of psychotic and bipolar history), and HPPD. Nutt's multicriteria harm analysis ranks it among the least harmful of assessed recreational drugs.26
The commercial landscape: psychedelics in Phase 2–3
Every psychedelic or psychedelic-adjacent compound in active late-stage development. Psilocybin's rows are highlighted; note that it appears three times, under three sponsors.
4-PO-DMT
same molecule
CYB003 / HLP003
lysergide D-tartrate
βk-MDMA
via luvesilocin
mebufotenin benzoate
mebufotenin
N,N-dimethyltryptamine
CYB004
midomafetamine
single enantiomer
How to read this table. Two structural facts stand out. First, almost nobody owns their molecule — psilocybin, LSD, DMT, 5-MeO-DMT, MDMA and 4-OH-DiPT are all unpatentable as compositions, so every programme competes on formulation, route, deuteration or prodrug chemistry. Second, the field has converged on shortening the session: intranasal, inhaled, buccal, subcutaneous and deuterated routes all exist to cut monitored clinic hours, which is the binding constraint on this entire class. MDMA is the cautionary entry — furthest along, and stalled on trial-conduct and blinding grounds rather than on biology. Psilocybin's own position is singular: three separate programmes (Compass, Usona, Cybin) are pursuing the same active moiety by different routes, which is why it will almost certainly be the first of this class to reach a US label.
Established, probable, speculative
Grading is ours. A — multiple independent sources, consistent. B — solid primary evidence, limited replication. C — single source, sponsor-derived, or contested.
Established · A
- Psilocybin is an inactive prodrug dephosphorylated to psilocin.3,5
- 5-HT2A agonism is necessary for the subjective effects — ketanserin blockade in humans.7
- Isolated by Hofmann in 1958 following Wasson's Oaxaca fieldwork; marketed as Indocybin; Schedule I in 1970.1,3
- A single 25 mg dose reduces depressive symptoms in TRD versus placebo and versus 1 mg.19,20
- Acute AEs are mild-to-moderate, dosing-day-clustered, mostly resolving within 24 h.21,23
Probable · B
- Benefit in responders persists through at least 26 weeks after one or two doses.20,22
- Two doses outperform one, and retreatment is well tolerated.20
- It reduces heavy drinking in alcohol use disorder.14
- It reduces depression and anxiety in life-threatening cancer, with 6-month persistence.12
- It promotes cortical structural plasticity via 5-HT2A.10
Speculative · C
- That the subjective experience is necessary rather than concurrent. The field's biggest open question.10
- That a 3.6–3.8-point MADRS difference is clinically sufficient for adoption and reimbursement.19,20
- The network-flexibility account of antidepressant response as mechanism rather than correlate.9
- Efficacy in OCD and smoking cessation.15
- How an approved product coexists with state supervised-use programmes.24
Research gaps, in order of consequence
Every operational problem in this field — session length, staffing, monitoring, cost — dissolves if 5-HT2A-mediated plasticity can be had without the acute experience. Psilocybin is the compound with enough data to answer it, and it has not been answered.
A 25–39% response rate with a modest mean difference implies real heterogeneity. No validated predictor of response exists. Finding one would transform the value proposition more than any effect-size improvement.
COMP006 shows two doses three weeks apart beat one. Nothing establishes that three weeks is optimal, how many doses are useful, or what happens on the third and fourth — which is precisely what chronic-disease management requires.
Support is present in every trial and separately characterised in none. Its contribution, minimum sufficient form, and practitioner-training requirements are all unquantified — the same problem that sank MDMA-assisted therapy.
Psilocybin's anxiety evidence comes from cancer-related distress, not from GAD. Neither Compass nor the academic groups have run a pivotal GAD trial — which is why LSD and 4-OH-DiPT have that space largely to themselves.
Trial exposure is hundreds of patients over months. HPPD incidence, psychosis risk in less-selected populations, and 5-HT2B exposure under repeated dosing all need post-approval surveillance to resolve.
Sources
P1 peer-reviewed primary research or official instrument · P2 review, book or historical scholarship · P3 sponsor communication or trade press · P4 general reference, non-load-bearing.