For most of the period after Lykos failed at the FDA in 2024, the binding constraint on psychedelic medicine was regulatory. The question was whether the agency would ever accept a psychedelic dossier at all. As of mid-2026, that question has been answered, and the constraint has moved. Compass Pathways has positive Phase 3 data, a rolling New Drug Application already underway, and a federal priority voucher that could shorten its review to a matter of weeks. The regulatory path is now the fast-moving part of the story. The slower part, and the part that will decide whether psilocybin becomes a real treatment or a narrow one, is the clinical data on how large the effect is and how long it lasts.
What the trials showed
On February 17, 2026, Compass reported that both pivotal Phase 3 trials of COMP360, its synthetic psilocybin formulation, met their primary endpoints in treatment-resistant depression. COMP005 compared a single 25-milligram dose against placebo and produced a 3.6-point separation on the Montgomery-Asberg Depression Rating Scale at week 6, with a p-value below 0.001. COMP006 compared two 25-milligram doses given three weeks apart against a 1-milligram comparator and produced a 3.8-point separation at the same timepoint, also below 0.001. This is the first time a classic serotonergic psychedelic has cleared Phase 3. The only psychedelic-adjacent product already approved is Johnson and Johnson’s esketamine, a ketamine derivative rather than a classic psychedelic.
That is the milestone, and it is a real one. The number that defines it is modest.
The effect-size question
A 3.6-to-3.8-point separation on the MADRS is statistically unambiguous and clinically small. The MADRS runs from 0 to 60, and the size of the drug-versus-comparator gap is the figure an FDA advisory committee will scrutinize, because it is the figure that translates into whether a patient and a prescriber would notice the difference. A separation in the three-to-four-point range is inside the territory where reasonable reviewers disagree about clinical meaningfulness, and it invites the same question that hangs over modern antidepressant and antipsychotic trials generally: how much of the measured gap is the drug, and how much is a function of the comparator and the trial design.
That question is sharper in psychedelic trials than almost anywhere else because of functional unblinding. A participant who receives an active 25-milligram dose of psilocybin generally knows it, and so, often, does the rater, which can inflate the apparent drug effect through expectation. COMP006’s use of a 1-milligram active comparator rather than inert placebo is a design choice aimed squarely at this problem, an attempt to give the control arm enough of a perceptible effect to blunt the guessing. Whether it succeeded is part of what reviewers will examine in the full datasets. The desk has written separately on why unblinding is the structural weakness of this trial class; COMP360 is the highest-stakes test yet of whether a well-designed psychedelic trial can hold up to that scrutiny.
The durability data is not all in
Compass markets COMP360 on rapid and durable response. The rapid half is demonstrated at week 6. The durable half, for the two-dose regimen, is not yet public. The 26-week data from Part B of COMP006 are expected in early in the third quarter of 2026. Until those land, the durability claim for the regimen most likely to reach the market rests on shorter follow-up and on the single-dose trial. For a treatment whose entire commercial logic is that a small number of sessions produces lasting benefit, the 26-week readout is not a footnote. It is the test of the thesis, and it arrives after the company intends to have most of its NDA filed.
The regulatory path moved fast
The regulatory progress since February has been unusually quick. In April 2026 the FDA granted Compass a rolling NDA submission and review, meaning sections of the application can be filed and reviewed as they are completed rather than all at once, and the company has said that process is underway. The agency also awarded COMP360 a Commissioner’s National Priority Voucher, a program whose benefits include enhanced communication and a target review window of one to two months after filing. Compass has said it expects to complete the final NDA submission in the fourth quarter of 2026, with launch readiness by year-end, and reported cash of $466 million as of the end of March, a runway it describes as reaching into 2028.
Rescheduling, the other federal gate, is also moving. A White House executive order on psychedelic treatments directs the DEA to review and, where warranted, reschedule psychedelic treatments that have completed Phase 3, so that a product cleared by the FDA would not then stall for lack of a controlled-substances pathway. Taken together, the rolling review, the voucher, and the rescheduling directive describe a federal apparatus moving to clear the path rather than block it.
Where the constraint now sits
This is the inversion worth marking. A year ago the open question in psychedelic medicine was whether the regulatory system would engage at all. Now the regulatory system is engaging quickly, and the open questions are clinical: whether a three-to-four-point MADRS separation is large enough to matter, whether it survives the unblinding scrutiny that sank confidence in earlier psychedelic datasets, and whether the two-dose benefit holds at 26 weeks. None of those are answered by an NDA being accepted for review. They are answered by the data inside it and by how an advisory committee reads that data.
The next real datapoint is the COMP006 26-week readout in the third quarter. The next institutional one is the advisory committee, if the FDA convenes one, where the effect size and the unblinding question will be argued in public. For a sector that spent two years asking whether the door would ever open, the better and harder question is now in front of it: whether what walks through is a broadly useful treatment or a narrow one for a difficult population. The regulatory speed does not settle that. The clinical data will.