MDMA-Assisted Therapy

← Encyclopedia

In June and August of 2024 a drug that the United States classes alongside heroin came within one regulatory step of being prescribed for one of psychiatry’s most stubborn conditions, and was turned away. The drug was MDMA, the compound sold on the street as ecstasy. The condition was post-traumatic stress disorder. The treatment under review was not a pill to be taken daily but something stranger to the pharmacy’s habits: a short course of psychotherapy in which the drug is given during a small number of supervised, day-long sessions, with non-drug preparation before and non-drug integration after. The molecule is held to be an adjunct, not the medicine. The therapy is the medicine, and the drug is what lets the patient stay in the room with what hurts.

That framing is not incidental. It is the reason the story bends the way it does. A drug can be regulated; a conversation cannot, at least not by the Food and Drug Administration, which approves compounds and not the talk that surrounds them. MDMA-assisted therapy asked an agency built to weigh molecules to weigh a hybrid — part chemistry, part relationship, part expectation — and the seams of that hybrid are where the case eventually came apart.

The pharmacology is the easy part to state, and the only part that can be stated at the level of survey. MDMA releases serotonin, raises oxytocin and related signaling, and is associated with a quieting of the fear response — a reduced reactivity, in the threat circuitry, to cues that would otherwise overwhelm. Patients describe trust, openness, a lowered guard. From this the therapeutic theory follows. Trauma work in ordinary therapy can stall because revisiting the memory pushes a person out of the range in which they can think and feel at once — into panic or into numb shutdown. Clinicians call that range the window of tolerance. The hypothesis is that MDMA holds the patient inside it: feeling enough to engage the memory, not so much as to flee it, so the memory can be approached and reworked rather than avoided. A related idea borrows the language of fear extinction from animal research: reactivate the memory with its terror dialed down, and its emotional charge may be updated when it settles. These are theories of why it might work. They are not yet established accounts of why it does, and the distinction will matter. Each rests on inference from what the drug does elsewhere — in healthy volunteers, in animal models of fear — rather than on a mechanism caught at work in a treated patient, and the trials, for reasons that become clear below, could not cleanly separate the pharmacology from the therapy or from what the patient expected of the day.

The history runs through a chemist’s curiosity. Merck first synthesized the compound in 1912, as a way station to something else, and it sat without use for sixty years. In the mid-1970s the American chemist Alexander Shulgin learned of its distinctive psychological effect, made it again, and in 1978 published the first account of how it felt in humans — “an easily controlled altered state of consciousness,” he wrote, “with emotional and sensual overtones.” Shulgin passed it to the psychotherapist Leo Zeff, who used it in his practice and carried it to others, and through the late 1970s and early 1980s a quiet therapeutic underground of clinicians employed it as exactly what the modern trials would later formalize: a means of lowering avoidance and strengthening the bond between patient and therapist. Then the drug escaped the clinic. As recreational ecstasy spread, the Drug Enforcement Administration moved it into Schedule I in 1985 and made the placement permanent the next year, over the objections of clinicians who valued it, and most American research stopped. The Multidisciplinary Association for Psychedelic Studies, founded in 1986, existed largely to keep that research alive through the long prohibition — and it was MAPS, through its corporate arm later renamed Lykos Therapeutics, that eventually brought the case to the FDA.

The case rested on two trials. Both were multi-site, randomized, double-blind, and placebo-controlled, both led by Jennifer Mitchell and published in Nature Medicine, and both compared the drug-plus-therapy against placebo-plus-the-same-therapy — an attempt to isolate what the molecule itself contributed. The first, in 2021, enrolled ninety adults with severe PTSD, a deliberately difficult group marked by dissociation, depression, and childhood trauma. On the standard clinician-rated severity scale the active group improved far more than the placebo group, a gap the authors reported as a large effect, and the trial recorded no signals of abuse, suicidality, or cardiac strain — a clean acute safety picture, on its own terms. The second, in 2023, was the confirmatory trial: a hundred and four adults across a broader and more diverse range of severity, and a similar result. The figures most often quoted — that roughly two in three of the treated dropped below the threshold for a PTSD diagnosis in the first trial, and seven in ten in the second, well above the placebo arms — come chiefly from an independent review and the sponsor’s own summaries rather than from each paper’s primary result, and are best attributed that way. The direction, though, was not in dispute. By its own numbers this was among the most promising developments in trauma treatment in a generation.

The numbers do not stand alone, and that is the whole of the second register. In June 2024 the FDA’s Psychopharmacologic Drugs Advisory Committee reviewed the application and voted against it — two to nine that the data did not show the treatment effective, one to ten that its benefits did not outweigh its risks. In August the agency issued a Complete Response Letter to Lykos, declining approval and asking for another Phase 3 trial. The reasons, made fully visible when the letter was released in 2025, were not minor. The central wound was functional unblinding: MDMA’s effects are so unmistakable that nearly everyone who received it knew, which means the placebo control did not control for the one thing it exists to control for — the patient’s belief that they have been given the real drug. An independent assessment by the Institute for Clinical and Economic Review had reached the same verdict that spring, calling the evidence insufficient and the trials “essentially unblinded,” and noting that strong prior convictions about the drug, held by participants and therapists alike, would have sharpened the bias rather than softened it. There were graver findings too. FDA inspections turned up adverse events that had gone unreported at multiple sites, and sites apparently trained not to log favorable reactions as adverse events at all, which cast doubt on whether the reassuring safety record could be trusted. ICER relayed reports that some therapists had encouraged positive accounts and discouraged negative ones. The samples were modest for a pivotal program, the durability of benefit past the trial window was thin, and the enrolled population may have been weighted toward people already disposed to respond. None of these is a fringe objection. Together they are the reason a treatment with striking topline results was not approved.

So the honest summary holds two facts at once. The effect on the page is real and large; its interpretation as a clean drug effect is exactly what remains unsettled. As of 2026 MDMA-assisted therapy is not approved in the United States, the path forward requires new trials, and the drug remains a Schedule I controlled substance — documented here, not advised, with no operative detail of any kind. What sits in the foreground is not a verdict but a question the field has not closed.

It is worth saying where this leaves the broader revival, because the difficulty that sank the application is not the drug’s alone. A trial earns its authority from the blind — from the fact that neither patient nor rater can tell who got what, so that hope and inert chemistry can be told apart. A substance whose subjective effects announce themselves breaks that machinery from the inside, and the same fault line runs through the clinical study of psilocybin and ayahuasca, where strongly felt experience and self-selected, hopeful volunteers leave expectancy free to move the scales. The MDMA case is the sharpest test the psychedelic renaissance has yet faced of its own method, and the most expensive failure of that test so far. Read against the placebo literature, the lesson is almost a definition: this is what happens to a placebo control when the active treatment cannot be hidden. The expectation the trial was built to subtract becomes inseparable from the thing it was built to measure.

There is an irony the field’s own history sharpens. The therapeutic underground that Shulgin and Zeff seeded prized MDMA for precisely the quality — the felt, unmistakable shift toward safety and openness — that the modern trial cannot quarantine. The thing that makes the drug useful in the room is the thing that makes it hard to prove in the data. A treatment can be felt to work, by patients and clinicians both, and still leave the question of whether it works, in the sense a regulator means, formally open. Between those two senses of the word lies the entire unfinished case, and the FDA’s answer, for now, was to ask again.

→ Related: Psilocybin Research · Ayahuasca Research · Placebo Effect · Transpersonal Psychology · Near Death Experience