What everyone believes

“Magic mushrooms (psilocybin) are overall good for your health”

The data's verdictmixed

“Psilocybin cures depression.” “Magic mushrooms are the safest recreational drug.” “It’s basically medicine now.” The last five years have flipped the popular story on mushrooms almost completely — from Schedule I menace to wellness breakthrough. The clinical data behind that flip is real. So is the fact that almost none of it describes how anyone actually eats mushrooms.

The belief

Magic mushrooms are, on balance, good for your health. It’s a claim that sounds unified but is actually three claims stacked on top of each other: that psilocybin helps people with diagnosed conditions, that this benefit shows up for people just trying it recreationally, and that the physical (not just mental) health picture is settled. Pull the stack apart and the evidence doesn’t move together.

0

MADRS points of depression-symptom reduction vs. placebo (p<0.001) — Compass Pathways' Phase 3 trial, 258 patients with treatment-resistant depression, single 25mg dose.

Where the belief is strongest: inside a clinic

Under screened, supervised, single-dose conditions, the case is genuinely good. Two Phase 3 trials have now hit their primary endpoint for treatment-resistant depression. The foundational 2016 Johns Hopkins/NYU trials found large, sustained reductions in depression and anxiety in cancer patients facing a terminal diagnosis. Early PTSD safety data is clean. And the effect sizes, where they exist, are larger than what standard antidepressants manage against placebo.

Depression effect size vs. placebo
Higher = bigger symptom improvement over placebo. The SSRI bar (Cipriani et al. 2018, 21-drug network meta-analysis) and the psilocybin bars (a separate meta-analysis) use related but not identical effect-size statistics — treat this as directional, not a strict apples-to-apples ratio. The high-dose psilocybin figure rests on a single trial; wide uncertainty, not yet replicated.

No new or unexpected safety findings; no clinically meaningful imbalance in suicidality between treatment arms.

— Independent Data Safety Monitoring Board, COMP005 trial

This is real, but it describes a specific thing: a screened participant — no personal or family history of psychosis-spectrum illness or bipolar I disorder — taking a controlled 25mg-range dose in a room with trained facilitators, with structured follow-up sessions afterward. Almost nobody eating mushrooms at a party is replicating that protocol.

Where the belief gets shakier: outside one

Two things happen once you leave the clinic. First, the studies get thinner. Second, the definition of “safe” starts doing a lot of quiet work.

Ten claims, two contexts — where 'good for your health' holds and where it doesn't
Split by where the evidence comes from (clinical/supervised vs. recreational/unsupervised) and by direction (supports the belief vs. counsels caution). The clinical column skews positive; the recreational column is genuinely split — real safety data alongside real, uncontrolled risk.

The 2017 Global Drug Survey — about 115,000 respondents across 50 countries — found magic mushrooms had the lowest rate of users needing emergency treatment among the recreational drugs surveyed.

0%

of ~10,000 surveyed magic mushroom users needed emergency medical treatment — versus roughly 5x that rate for alcohol, MDMA, and LSD users in the same survey.

That statistic is real and it’s the headline “safest drug” claim gets built on. But it measures one narrow thing — the rate of a bad-enough reaction to need an ER — not general health impact, and it says nothing about who’s at elevated risk. The same survey found 11% of unsupervised users put themselves or someone else at risk of physical harm, concentrated in exactly the conditions clinical protocols exist to prevent: poor mindset, poor setting, mixing with other substances.

Adverse events per exposure, by definition and context
Same drug, three very different rates — because 'adverse event' means three different things here. Oregon's official definition only counts events needing emergency/medical/hospital response; its own detailed self-report follow-up, using a symptom checklist, found a rate roughly 4x higher at one day post-session (though those symptoms had resolved by 30 days for every case). The recreational figure uses the Global Drug Survey's 'needed emergency treatment' self-report.

The gap between 5.2 and 23 per 1,000 isn’t a contradiction — it’s the same Oregon program measured two different ways, and it’s a useful warning about any single “adverse event rate” claim in this space: ask which definition produced it before comparing it to anything else.

The physical-health picture is younger and less settled

Nearly everything above is mental health. The physical-health story is thinner, newer, and currently points toward caution rather than benefit. Psilocin (psilocybin’s active metabolite) binds the 5-HT2B receptor — the same receptor whose chronic overstimulation caused the withdrawn diet drug fenfluramine and the Parkinson’s drug pergolide to damage heart valves. No confirmed human case of psilocybin-linked valvulopathy exists yet, and the concern is specifically tied to repeated, chronic exposure — the pattern daily or near-daily microdosing produces, not a single high-dose session.

There’s also a sharp, unambiguous interaction risk that has nothing to do with dose or setting.

0%

of documented lithium + classic-psychedelic case reports involved a seizure — versus zero in the comparison group taking lamotrigine, another mood stabilizer, alongside the same drugs.

Seizure rate by co-administered mood stabilizer
62 lithium case reports vs. 34 lamotrigine case reports, drawn from an analysis of online psychedelic experience reports. Lithium is treated as a hard contraindication in every clinical psilocybin protocol for exactly this reason.

Where the law sits, and why that’s not a health signal

Psilocybin remains Schedule I federally — “no currently accepted medical use” in DEA terms — even as the FDA has granted breakthrough therapy status to three separate psilocybin programs and a 2026 executive order directs the FDA and DEA to accelerate research and patient-access pathways. Oregon and Colorado run state-regulated programs that operate legally alongside that federal schedule. None of this is a health verdict by itself: the regulated programs screen out the same high-risk individuals the clinical trials do, and “legal to consume at a licensed center” says nothing about safety for someone using mushrooms outside that structure.

Verdict

Mixed. The belief is well-supported in exactly the setting almost no one associates it with — a clinic — and considerably less settled everywhere else.

For a specific, diagnosed condition (treatment-resistant depression, end-of-life anxiety, and increasingly PTSD), under screening that excludes people with a personal or family history of psychosis-spectrum or bipolar I illness, with a controlled dose and trained support: the trial evidence is real, if still methodologically young. That’s a genuinely good-for-your-health story, and it’s the one driving the FDA pipeline.

Strip away the screening, the dose control, and the supervision — which is how most people who’ve ever tried magic mushrooms actually took them — and the picture splits. The aggregate safety numbers are still comparatively good (lowest ER-visit rate among recreational drugs surveyed). But that aggregate hides real risk for a subset of people (the psychosis/bipolar-vulnerable group entirely excluded from the positive trial data), a hard and specific danger for one common combination (lithium), a physical-health question that’s barely been asked yet (chronic cardiac exposure), and an adverse-event rate that quadruples depending on which definition of “adverse event” you use.

So: not a wonder drug, not a scare story. A drug whose benefit is currently proven for narrow, supervised use — and whose risk, for the much broader population that actually uses it recreationally, is real but concentrated in identifiable places rather than spread evenly across everyone.