The FDA Wrote the Rulebook for Psychedelic Trials. It Is Not a Green Light.

On July 14, 2026, the Food and Drug Administration published its final guidance on psychedelic drug development in the Federal Register. The document, “Psychedelic Drugs: Considerations for Clinical Investigations”, comes from the Center for Drug Evaluation and Research and finalizes a draft that had been circulating since June 2023. In the same notice, the agency announced a public hearing on the therapeutic use of psychedelics, which it held on September 14.

Coverage has largely framed the guidance as a signal of regulatory warmth — arriving as it does after Executive Order 14401 in April, which directed federal agencies to accelerate psychedelic research, and alongside National Priority Vouchers that compress FDA review windows for Compass Pathways, the Usona Institute, and Transcend Therapeutics. Analysts read it as de-risking the sector.

That reading is not wrong, but it skips what the document actually says. Every page carries the standard header: “Contains Nonbinding Recommendations”. In its background section the agency states that psychedelic drug development programs “are subject to the same regulations and same evidentiary standards for approval as other drug development programs”. There is no abbreviated pathway and no special category. What there is instead is unusually specific: a discipline-by-discipline account of where this class of compounds tends to generate uninterpretable data, and what the agency would accept as a response.

What the guidance actually says

The core of it addresses functional unblinding — the problem that a participant given an active dose of psilocybin, LSD, or MDMA knows it, and so does the clinician watching. The perceptual effects are not a side effect to be managed around; they are the mechanism. In a trial whose primary outcome is a clinician-administered symptom rating, that mutual knowledge contaminates the measurement at both ends. The guidance spells out the mechanism: functional unblinding produces expectation bias in patients who experience perceptual disturbances and in those who observe them, and the reverse in those who get placebo and notice nothing.

The scale of the problem is not hypothetical. A 2023 ACTTION systematic review found that of 86 randomized psychedelic trials reporting blinding procedures, only eight actually assessed whether the blind had held.

The countermeasures are recommendations, not requirements — the document uses “should” throughout, and says so explicitly. Sponsors may use comparators other than inert placebo, including lower doses of a psychedelic or other psychoactive drugs that mimic part of the subjective experience. Blinding questionnaires for both subjects and investigators — asking each to guess the allocation and rate their confidence — are described as helpful for evaluating how far unblinding occurred. An expectancy questionnaire before randomization and at the end of treatment “should be considered”. For trial architecture, the agency suggests pairing a conventional placebo-controlled study with a separate dose-response study using low, middle, and high doses and no placebo, so the dose-response relationship can be characterized with less risk of unblinding while the placebo-controlled arm does the work of characterizing safety.

Then it sets the bar for what survives. Study results, the guidance says, “should be strongly persuasive and robust across study endpoints to overcome biases that may be introduced by functional unblinding”. That sentence is the whole document in miniature: no lowered threshold, an explicitly raised one.

The safety expectations are similarly concrete. The agency expects each treatment session to be observed by two monitors for its full duration — a lead monitor who is a licensed, independently practicing clinician with graduate-level psychotherapy training, and an assistant with a nursing or bachelor's degree and at least a year in a licensed mental health setting. If the lead monitor is not a physician, a licensed on-call physician must be able to reach the site within fifteen minutes. Drugs with functional activity at the 5-HT2B receptor — a subtype the guidance links to heart valvulopathy — call for baseline and follow-up echocardiograms when chronically administered, with patients who have preexisting valvular disease or pulmonary hypertension excluded until the risk is characterized.

And there is a provision that reads strangely until you understand why it is there: expected psychoactive effects, including euphoria, hallucinations and other perceptual distortions, and alterations in cognition, should be recorded as adverse events “even if subjects do not describe these effects as adverse”. The therapeutic experience, logged formally as a potential harm — not because the agency thinks a good trip is an injury, but because those effects are the signal of abuse potential, and abuse potential is assessed separately from therapeutic benefit.

What it does not settle

The guidance's most consequential passage is an admission. As of its publication date, the FDA writes, “the contribution of the psychotherapy component to any efficacy observed with psychedelic drug treatment has not been characterized.”

Nearly every late-stage psychedelic program was built on the integrated model MAPS established for MDMA: drug administration wrapped in preparation, in-session support, and integration. The agency is saying that after three decades of modern research, nobody has demonstrated what the therapy half contributes. Its proposed remedy is a factorial design separating the two, with an operational suggestion attached — one option for limiting bias, it notes, is to ensure the in-session monitor is not the person delivering post-session psychotherapy, since a monitor can usually deduce allocation from behaviour and bias everything downstream.

That question has consequences well beyond study design. If a sponsor shows the molecule works independently, the label need not mandate intensive psychological support, and administration could eventually look like ordinary outpatient prescribing. If the evidence shows the drug only works inside the therapy, the treatment paradigm used in the trials may be written into the label — making the expensive, supervised, two-monitor session model the only lawful way to receive it. The guidance says as much, in one flat sentence: the treatment paradigm used in the trials may be described in product labeling.

The agency did not lower the bar for psychedelics; it drew a detailed map of where the bar sits, which is a harder favour than the one the field was hoping for.

What the September hearing was actually about

The public hearing the agency held on September 14 was not about whether psychedelics work. It was about what happens if one is approved — specifically, who would be permitted to administer it, and in what setting. Reporting from the hearing describes access as the dominant theme, with most speakers pressing the FDA to ensure the widest possible patient access and a smaller number raising safety concerns.

That is the guidance's unresolved question surfacing in a different room. The two-monitor, full-session model is expensive and clinician-intensive. If it becomes a condition of the label, supply is constrained by the number of trained clinicians willing to sit through long dosing sessions. If it does not, the safety architecture the guidance describes exists only inside trials. Neither the hearing nor the guidance resolves this, and the agency has given no date by which it will.

Where this leaves the approval timeline

Compass Pathways is furthest along. Its synthetic psilocybin, COMP360, produced positive primary results in two consecutive Phase 3 trials — COMP005 in June 2025 and COMP006 in February 2026, both against a 1 mg comparator — and reported 26-week durability data from COMP006 on July 7. Under Breakthrough Therapy Designation the company is submitting a rolling NDA, targeting completion in the fourth quarter of 2026, with the priority voucher compressing review to roughly one to two months after a complete submission. Commissioner Marty Makary has said he expects an approval decision on at least one priority compound by late summer or fall.

Two caveats are worth holding onto. A commissioner's stated expectation is a prediction, not a commitment, and review timelines slip routinely. And FDA approval alone would not put psilocybin in a pharmacy: psilocybin is a Schedule I substance, and the guidance notes that approval would trigger a rescheduling action under the Controlled Substances Act — a separate DEA proceeding that does not follow automatically. Executive Order 14401 directs the Attorney General to review scheduling for any psychedelic completing Phase 3 and to act as quickly as practicable, but sets no deadline. Compass lists DEA rescheduling as a risk factor in its own SEC filings.

What a reader should take from this

The regulatory environment for psychedelics has genuinely changed, and the change is real enough that a first FDA approval in this class is now plausible within the next year rather than hypothetical. But the guidance is not a relaxation of standards. It is the opposite: a written, detailed account of why most psychedelic trials to date cannot support the conclusions drawn from them, and what a trial would have to look like to do better. The single largest open question in the field — whether the drug or the therapy is doing the work — remains, by the agency's own statement, uncharacterized. Anyone reading the July guidance as vindication of psychedelic medicine is reading a document that says something closer to the reverse: the science has been permitted to proceed, on condition that it become considerably more rigorous than it has been.

This article is general information, not medical advice. See our Editorial Standards.