327 registered trials. Most name one condition, some name several, which is why the rows add to more than 327. Seventeen fall outside these groups.

The second row surprises people, and it needs care. Sixty-six trials give psilocybin to healthy volunteers to observe what it does in the body. Fourteen of those also name a patient condition, eight of them depression, so the number that names no patient condition at all is 52 of 327. A healthy-volunteer study measures how a drug behaves. It is not a study of whether it treats anything.

Every number here counts a registration: a sponsor filing a protocol on a public registry saying what it intends to study. Three things follow.

A registration is not evidence. Some of these trials have finished and published, some are still recruiting, and some will never report. Counting them measures where money and attention went. It measures nothing about outcomes.

A registry is not the whole of science. Laboratory work and individual case reports sit outside it by design. When a group in the table above holds few trials, that means few registered human protocols exist. That is a fact about the register, and it carries no verdict on the idea.

And these are counts of trials, never of patients. The smallest study in the register enrolled four people and the largest a thousand. Each is one row.

Then the honest answer is a small number, and it is worth saying why.

Anxiety holds 18 registrations of 327. OCD holds 12, eating disorders 8, bipolar disorder 4. Those totals are small enough that two or three new filings would visibly change the picture, which is why we will not tell you which way any of them is moving. It is not that we have a view we are keeping back. At those numbers there is nothing yet to see.

Your condition may be in the register even when this table does not show it. Seventeen trials match none of our groups, and their own condition strings are more varied than the eleven suggest: post-treatment Lyme disease, irritable bowel syndrome, burnout in physicians, caregiver distress, aging and longevity, self-harm, stable ischaemic heart disease. Two of the seventeen name conditions our groups do hold, which is a gap in our matching and one we are repairing. We drew the eleven groups. Medicine did not. Search the register by your own condition and treat this table as a summary.

If your next question is whether any of this is open to you, the trials in the recruiting stage are listed separately.

It did not. It grew around it.

Split the register by start date: 99 trials are dated 2022 or earlier and 222 are dated 2023 or later, leaving six with no start date and out of the comparison. Four groups grew faster across that split than depression did. Neurological conditions went from 3 to 18, PTSD from 6 to 24, addiction and substance use from 12 to 40, and cancer and end-of-life care from 7 to 18.

Depression went from 35 to 82. Every other trial in the register, taken together, went from 64 to 140. Depression grew by a factor of 2.34, the rest by 2.19: the largest category in the field expanded slightly faster than all the others combined. Work on healthy volunteers grew more slowly than either, from 23 to 43.

by start date, trials dated 2022 or earlier against 2023 or later · six trials carry no start date and are excluded · five of the later trials have start dates still in the future, one of them 2029

The register here is our own corpus, which merges three public trial registries and holds 327 rows. Groups come from our condition taxonomy, condition_groups v2026-08-20, applied to all of them. A trial appears in every group it matches, so the eleven groups hold 384 assignments over 327 trials, and the table above totals 401 once the seventeen ungrouped are added. Counts are of trials.

Trials now name more conditions each, 1.20 groups per trial among those dated 2023 or later against 1.12 among the earlier ones. That is part of why every category looks larger, and it is why this piece compares growth rates. Shares of trials do not sum to 100 here, so no group’s rise takes anything from another. Depression holds 35.4% of the earlier trials and 36.9% of the later ones, a rise of 1.6 points; measured instead as a share of group assignments, the only denominator that does sum to 100, the same move becomes a fall of 0.8. Both are printed because neither is the story; the growth rates are. Depression is compared with the rest of the register. Comparing it with the register as a whole would compare it partly with itself.

Two readings cut against ours. Composition is not merit: the fastest-growing categories are the ones carrying large public funding streams, so this may be a map of grant money that is only incidentally a map of the science. And small starting cells make unstable factors: neurological conditions grew sixfold, but that is three trials becoming eighteen. Depression’s own margin is narrow, 2.34 against 2.19, and a handful of rows either way would close it.

The corpus moves, so every figure here is re-derived before publication and the script fails loudly if any has changed.