What federal policy actually changed in 2026

Why “fast-tracked” keeps getting misread

When NBC News ran a segment titled “FDA grants quick review for 3 psychedelic drug trials,” the reporting was accurate. What happened next was not: within days, the story had been compressed in a thousand social posts into “the FDA is approving psilocybin.”

A Commissioner’s National Priority Voucher shortens the administrative clock. Instead of the agency taking ten to twelve months to work through an application, it takes one to two. That is genuinely significant — it could compress the timeline by the better part of a year. But it operates entirely on the calendar, not on the standard. The sponsor still has to demonstrate safety and efficacy to the same bar as any other drug. A voucher is a place in the queue, not a verdict.

The two events everything waits on

This is the mechanism worth understanding, because it explains why so many states appear to have legalized psilocybin without anyone being able to get it.

1. The FDA approves a specific psilocybin medicine for a specific indication. Not psilocybin generally — a formulation, with a label.
2. The DEA reschedules that formulation out of Schedule I, which is what makes it prescribable.

4 states — Arizona, Georgia, South Dakota, Virginia — have already written laws that activate automatically when both occur. Until then those laws authorize nothing at all. A reader in Virginia who has seen “Virginia legalizes psilocybin” is reading about a statute that currently does nothing.

See the state-by-state tracker for where each state actually stands.

What SAMHSA told the states, and why it matters

The August 2026 report is the most consequential document of the year for anyone who has to plan rather than opine. It tells states to build credentialing standards, safety protocols and complaint mechanisms before approval, and it names a budget problem nobody has solved.

The problem is this. Psilocybin therapy is mostly labor — a six-to-eight-hour session with a trained facilitator present throughout, bracketed by preparation and integration appointments. A state could cover the drug, or the therapy, but the report is blunt that prescribing psilocybin for unsupervised home use is not supported by the current evidence. So covering the drug alone is not a real option, and covering both is expensive at scale.

That is not an abstraction. It is the reason a legal psilocybin market could arrive and still be unaffordable to most of the people it was justified by — a question we track on the cost page.

The evidence, at the same standard we apply to everything else

Federal momentum is not evidence. It is worth stating plainly what the underlying data show, because the policy story and the science story are moving at different speeds.

Compass’s two Phase 3 trials both met their primary endpoints — COMP005 with a 3.6-point MADRS advantage over placebo at week six, COMP006 with 3.8 points for the 25 mg dose against 1 mg.^[2] Those results are real and statistically significant. They are also modest, and in the same broad range as conventional antidepressants in acute trials. We are not going to criticize antidepressants for a moderate average effect and then wave that through.

And the field’s most-quoted result is its most misrepresented: in the 2021 trial comparing psilocybin with escitalopram, the primary outcome was not statistically significant (p=0.17). The famous response and remission figures are unadjusted secondary outcomes, and the authors state explicitly that no clinical conclusions can be drawn from them.^[3]

The risks side of the ledger

Johns Hopkins gives “psilocybin research and potential” and “psychedelic risks, side effects and potential harms” equal billing on the same page. That is the right instinct, and rarer than it should be. So, briefly:

• Lithium is a genuine contraindication — the combination is associated with seizures.^[4] MAOIs likewise. Antipsychotics will block the experience entirely.
• Personal or family history of psychosis or bipolar I is an exclusion in essentially every trial, for good reason.
• Blinding is a real methodological problem. Participants and investigators can almost always tell who received the drug, which inflates apparent effects in ways trials rarely measure. The VA names this explicitly in its own assessment.^[5]
• Trial results mostly come from unmedicated people. Most psilocybin trials required participants to be off antidepressants before dosing. If you are currently taking one, you are not the population those numbers describe.

What we are watching next

• Compass’s rolling NDA submission, targeted for Q4 2026.
• Whether the Attorney General’s Schedule I review produces anything concrete.
• New Mexico’s program, targeting first patient access by end of 2026.
• Massachusetts H.5562, which passed the House 148–2 in July 2026 and has not become law.
• Whether any state publishes an actual Medicaid coverage framework in response to SAMHSA.