Anorexia nervosa carries one of the highest mortality rates of any psychiatric diagnosis and has almost no approved pharmacological treatment. A pilot study from Imperial College London’s Centre for Psychedelic Research, years in the making and now published in the British Journal of Psychiatry, reports that psilocybin therapy produced sustained improvement in eating-disorder symptoms in a small group of women who had already failed every prior treatment available to them. That is a real and clinically meaningful signal in an indication where the bar for any positive finding is low, precisely because so little else works. It is also a study with no control group at all, and reading the result honestly means holding both facts at once.
The trial
Twenty-one women with anorexia nervosa, averaging eleven years living with the illness and a body mass index of 16.4, well below the healthy threshold of 18.5, and none of whom had found lasting success with prior treatment, received three doses of oral psilocybin over six weeks alongside extensive talk therapy before and after each session. The first dose was 1 milligram, intended as a functionally inactive comparator, followed by two 25 milligram doses two weeks apart. Participants knew they were receiving psilocybin throughout; the only thing withheld was which specific dose they were getting at each session. There was no placebo arm and no comparison group of any kind. This is, by design, an open-label feasibility and preliminary-outcome study, not an efficacy trial.
What was found
Eating-disorder symptom severity scores improved at every follow-up point measured, two weeks, three months, six months, and one year after the final dosing session. Eighteen of the twenty-one participants showed improvement at two weeks. By three months, 48 percent scored in a range comparable to someone without an eating disorder. Participants also reported sustained increases in motivation to recover, holding through the full year of follow-up. Retention was notably strong for this population, twenty of twenty-one participants completed the dosing protocol, a population where trial dropout is typically a major obstacle. Body mass index showed no meaningful change, though the researchers note this was tracked only across the six-week dosing window, too short a period to expect a physical outcome like weight to shift meaningfully even if the psychological intervention is working.
Why the design sets a hard ceiling on the claim
Every result above has to be read through the absence of a control group. There is no comparator arm showing what these same twenty-one women, or a matched group with the same illness duration and severity, would have reported over the same year without psilocybin. Motivation to recover, self-reported symptom severity, and general engagement with treatment can all move simply from the experience of receiving sustained, intensive clinical attention, the extensive therapy accompanying each dose, the novelty of a new intervention after years of treatment failure, and the expectancy effects this desk has tracked across the psychedelic field generally. None of that means the improvement observed here is illusory. It means this design cannot distinguish a genuine psilocybin-specific effect from a strong general treatment-and-attention effect, and the researchers themselves are explicit that this is a preliminary result requiring larger, controlled studies before any causal claim is warranted.
The trial’s own low-dose comparator illustrates the field’s recurring blinding problem in miniature. The 1 milligram first dose was intended to function as an inactive placebo condition within the protocol. Some participants reported noticeable effects from it anyway. When even the built-in low-dose control cannot be reliably distinguished from an active dose by the people receiving it, it underscores how difficult genuine blinding is to achieve with this class of drug, a problem that shows up here even inside a single-arm study with no separate placebo group to begin with.
The safety record, reported plainly
The most common adverse effects were headache and nausea, and one participant withdrew from the study after her second dose. One participant experienced a serious adverse event during the follow-up period, unrelated in timing to the dosing sessions themselves, involving hospitalization following suicide attempts at seven and nine months after her final dose. Imperial’s research governance and clinical teams reviewed the case and assessed it as not related to the psilocybin intervention. That assessment should be read in context rather than taken as the final word: anorexia nervosa carries a high baseline rate of suicidality independent of any treatment received, which is part of why the assessment concluded the event was unrelated, but a single serious event in a sample of twenty-one is also too small a number to draw a confident conclusion about causality in either direction. The honest position is that the event occurred, it was reviewed by the appropriate institutional bodies, and the sample size here cannot settle the question definitively.
Why the researchers’ own framing is worth taking seriously
The senior author, psychedelic researcher Robin Carhart-Harris, and the study’s lead author both explicitly describe this as a pioneering but preliminary result that justifies larger, more rigorous studies, not as a demonstrated treatment. That is the correct register for what a 21-person, open-label, single-sex, uncontrolled pilot can support, and it is worth noting when a research team states the limitations of its own work this clearly rather than reaching past the data. The eating-disorder specialist who led therapy on the trial framed the value in more modest, mechanistic terms, that the intervention gave participants a different way to examine their relationship to a fiercely protective, often deeply entrenched illness, language that reads as a hypothesis worth testing further rather than a result to build a treatment protocol on today.
The frame
This piece is a genuine and carefully conducted addition to a very short list of things that have shown any signal at all in treatment-resistant anorexia nervosa, and the retention rate and directional symptom improvement are meaningful in a population where both are usually hard to obtain. It is also, unavoidably, a study that cannot separate its own intervention from the effects of intensive attention and expectancy, because it was never designed to. The field’s pattern holds here just as it has across psilocybin’s other indications: real, hard-won pilot signal, immediately followed by the harder and slower work of building a trial that can actually test whether the drug, rather than everything surrounding it, is doing the work. For an illness this dangerous and this resistant to treatment, that slower work is worth doing carefully rather than skipping past.
This piece discusses a serious mental health event, including a suicide attempt, in the context of clinical trial safety reporting. If you or someone you know is struggling, the 988 Suicide and Crisis Lifeline is available by call or text at 988 in the United States.