Cannabis & Mental Health: What 28 Clinical Trials Actually Show
A 2026 scoping review of 28 RCTs across 12 DSM-5 conditions. Where cannabinoids show benefit, where evidence is thin, and what we still don't know.
Professor High
The Honest Headline
May is Mental Health Awareness Month, and cannabis social media right now is full of confident claims: cannabis treats depression, cures anxiety, fixes PTSD. Often the post is selling something. Honesty is more useful than advocacy: cannabis is not a mental health miracle. The science is more nuanced — and much thinner than the marketing suggests.
In December 2025, Sophie Cooling and colleagues at the University of Melbourne published a scoping review in Clinical Drug Investigation synthesizing every RCT of medicinal cannabis for DSM-5 mental health conditions. From 8,061 candidate studies, they ended with 28 trials covering 12 conditions — the entire global body of gold-standard evidence for cannabis as primary psychiatric treatment.
In March 2026, a separate team published a meta-analysis in The Lancet Psychiatry covering 54 RCTs through May 2025. The reviews tell a consistent story: short-term signal in a few places, a great deal of nothing in many others, and essentially zero evidence for sustained long-term efficacy.
This isn’t a takedown. I run a cannabis intelligence platform and think cannabis has real value in many people’s lives. But the difference between “this plant helps me get through the day” and “this plant is clinically validated medicine for my diagnosis” is enormous.
A note before we proceed: This is a science explainer, not medical advice. Do not start, stop, or substitute any psychiatric treatment based on a blog post. If you’re considering cannabinoids for a diagnosis, that’s a conversation for a prescribing clinician — ideally one informed about cannabinoid pharmacology and not financially dependent on selling you cannabis.
What the Scoping Review Actually Did
A scoping review maps the territory rather than pooling numbers into an effect-size estimate. The Cooling team chose this format because the underlying trials were too heterogeneous — too different in product, dose, duration, and outcome — to honestly pool.
They searched PubMed, Web of Science, and PsycINFO for RCTs from 1980–2024 testing medicinal cannabis (CBD, THC, nabiximols, smoked or vaporized flower) for a primary DSM-5 diagnosis. From 8,061 records, 28 trials met inclusion criteria across 12 countries (most often the US, Canada, and Australia). Sample sizes ranged from 6 to 150 participants, median 42. Follow-up ranged from a single day to 13 weeks, median 6 weeks.
Read those numbers again. The median trial enrolled 42 people and followed them for six weeks. Mental health conditions are usually chronic. Six weeks tells you almost nothing about whether something keeps working over the months and years patients actually need.
The 12 Conditions Covered
Here is the breakdown of how the 28 RCTs distributed across diagnoses:
- Schizophrenia / psychotic disorders — 5 trials
- Cannabis use disorder (CUD) — 4 trials
- Cocaine use disorder — 4 trials
- Post-traumatic stress disorder (PTSD) — 3 trials
- Anxiety disorders — 3 trials
- Opioid use disorder — 2 trials
- Autism spectrum disorder (ASD) — 2 trials
- Major depressive disorder — 1 trial
- ADHD — 1 trial
- Obsessive-compulsive disorder (OCD) — 1 trial
- Tobacco use disorder — 1 trial
- Tourette syndrome — 1 trial
What’s striking: depression — the most prescribed-for mental health condition in the world, and one of the most common reasons people self-medicate with cannabis — has exactly one RCT. ADHD, OCD, and Tourette each have one. The 2026 Lancet Psychiatry meta-analysis went further and found no RCTs evaluating cannabinoids for depression as a primary indication at all. Substance use disorders make up nearly half the entire literature.
Where Cannabinoids Showed Real Promise
The Cooling review identified three areas with the most consistent short-term signal: cannabis use disorder, autism spectrum disorder, and schizophrenia (specifically as adjunctive CBD). Let’s take each.
Cannabis Use Disorder: The Paradox
The strongest, most replicated finding is that balanced CBD/THC products may help people reduce problematic cannabis use. The 2026 Lancet Psychiatry meta-analysis found CBD/THC combinations significantly reduced withdrawal symptoms (SMD −0.29) and weekly grams consumed (SMD −1.00) versus placebo.
Mechanistically this is coherent: heavy THC withdrawal produces irritability, sleep disruption, anxiety, and craving, and a controlled cannabinoid product can blunt those signals during a taper, similar to nicotine replacement for tobacco cessation. The drug isn’t curing anything — it’s smoothing a transition. (And CHS and CUD itself are real, common consequences of frequent high-potency use that the industry has not been forthright about.)
Autism Spectrum Disorder
Two RCTs in Cooling, plus a broader pediatric literature, suggest that CBD-rich products may reduce certain refractory behavioral symptoms in children with autism — aggression, self-injury, severe sleep disruption. The Lancet Psychiatry meta-analysis reported a modest reduction in autistic traits (SMD −0.36).
Two caveats. This isn’t “cannabis cures autism” — it’s “in children with severe, treatment-resistant behavioral symptoms, cannabinoids produced measurable short-term improvement on certain scales.” Second, evidence quality was rated low and trial sizes were small. Deeper dive: cannabis and autism research.
Schizophrenia: CBD as Adjunct
Schizophrenia was the most-studied condition (5 trials). Purified CBD as an adjunct to standard antipsychotic medication showed modest improvements in some symptom domains, particularly positive symptoms. CBD acts through mechanisms distinct from dopamine D2 antagonism, which makes the adjunctive idea pharmacologically appealing.
Critically, this is not a finding that cannabis as commonly used helps psychotic disorders. The opposite is well-documented: high-THC cannabis is associated with elevated psychosis risk and worse outcomes in established schizophrenia. The 2019 Black et al. meta-analysis found pharmaceutical THC worsened negative symptoms. See Cannabis and Schizophrenia Risk.
CBD ≠ THC ≠ cannabis flower. The schizophrenia signal applies narrowly to purified, high-dose CBD as an adjunct under clinical supervision — not dispensary flower.
Where the Evidence Is Thin or Mixed
Now the harder part of the conversation.
Depression
The elephant in the room. Depression is one of the top reasons people reach for cannabis, and our science-backed depression strain guide is one of our most-read articles. And yet: Cooling found one RCT for primary depression. The 2026 Lancet Psychiatry meta-analysis found zero. The 2019 Black et al. review found “scarce evidence” of benefit.
This isn’t “cannabis doesn’t help depression.” It’s “we haven’t done the studies.” Many people report cannabis improves mood short-term; others report it flattens them or worsens anhedonia, with depressive episodes intensifying during tolerance breaks. Both can be true. The rigorous comparative evidence doesn’t yet exist. If cannabis helps your depression, that’s a lived experience worth respecting — and not interchangeable with evidence-based antidepressant treatment.
Anxiety
Three RCTs in Cooling. The 2026 Lancet Psychiatry meta-analysis found no significant effect of cannabinoids on anxiety versus placebo. The 2019 Black et al. review found “very low quality” evidence of small benefit, and only in patients taking cannabis for other conditions (chronic pain, MS) — not primary anxiety.
Higher-quality work on purified CBD for acute anxiety (situational or experimental) shows more consistent benefit at 300–600 mg, but isn’t well-replicated in chronic anxiety. Many of the most-cited positive findings come from a handful of researchers and haven’t been reproduced at scale.
Pragmatic reality: many people use cannabis to take the edge off, and our guides on social anxiety strains and stress-relief strains reflect what works in practice. Terpenes like linalool, limonene, myrcene, and caryophyllene show plausible anxiolytic mechanisms preclinically. But “plausible mechanism plus user reports” isn’t the same evidentiary tier as clinical trial data.
THC is also famously biphasic for anxiety: low doses tend to be anxiolytic; higher doses, particularly in inexperienced users, often cause anxiety and panic. Balance products with substantial CBD generally feel safer for anxiety-prone users than high-THC Uplift products.
PTSD, ADHD, OCD, Tourette
PTSD: three RCTs, mixed, no significant overall effect in the 2026 meta-analysis. Observational data on nabilone for nightmares is the most-cited positive signal. Deeper dive: Cannabis and PTSD.
ADHD and OCD: a single RCT each. Insufficient to draw conclusions.
Tourette: one RCT in Cooling, but the 2026 meta-analysis identified a meaningful reduction in tic severity (SMD −0.68) — one of the more robust positive findings in the literature.
The Long-Term Question Nobody Wants to Talk About
The part of the Cooling review that should land hardest:
“no trial demonstrated long-term efficacy.”
Across 28 RCTs, 12 conditions, and four-plus decades of research, the longest follow-up was 13 weeks. The median was six. We have essentially zero rigorous data on whether cannabinoids continue to help beyond about three months.
Mental health treatment is a long game. SSRIs are studied across 6-, 12-, and 24-month windows because that’s the timescale on which depression and anxiety exist. Tolerance, dependence, dose escalation, and withdrawal effects emerge over months. Observational data show heavy long-term cannabis use can produce its own psychiatric burden — CUD, anhedonia, cannabis-induced anxiety, elevated psychosis risk in vulnerable populations. Short-term trials don’t capture these.
What This Means In Real Life
The single most important framing in this article:
“Recreational use that helps my mood” is not the same evidentiary claim as “cannabinoid as therapy.” Both can be true. They are different things.
If you smoke a joint after a hard week and feel better, that’s real. So is a glass of wine, a long run, or a hot bath. None of these have RCT evidence as treatment for a DSM-5 disorder, and none of them need to. The problem starts when “cannabis helps me cope” drifts into “cannabis is medicine, so I don’t need other treatment.”
The Cooling review and the Lancet Psychiatry meta-analysis aren’t telling you cannabis is worthless. They’re telling you the gap between marketing and clinical evidence is enormous, and serious psychiatric conditions deserve the higher bar.
The Risks Side of the Ledger
The 2026 Lancet Psychiatry meta-analysis found significantly increased all-cause adverse events in cannabinoid groups versus placebo (OR 1.75; number needed to harm = 7). The 2019 Black et al. review found roughly doubled adverse event rates and tripled treatment-withdrawal rates. Risks worth weighing:
- Psychosis risk — Heavy cannabis use, particularly high-THC use before age 25, is consistently associated with increased risk of new-onset psychosis in vulnerable individuals. One of the most robust epidemiological signals in cannabis science.
- Cognitive effects — Particularly in adolescents, whose endocannabinoid system is still developing. Persistent heavy use is associated with measurable, partially-reversible changes.
- Cannabinoid hyperemesis syndrome (CHS) — Increasingly common with heavy long-term use. See our CHS guide.
- Cannabis-induced anxiety — Paradoxical worsening, particularly with high-THC products at high doses in inexperienced users.
- Cannabis use disorder — About 9% of users develop dependence; the figure rises with frequency, age of initiation, and potency.
- Drug interactions — CBD interacts with many psychiatric medications via cytochrome P450 enzymes. Your prescriber needs to know.
None of this makes cannabis dangerous like alcohol or opioids — comparative harm data still places it on the lower-harm end of psychoactive substances. But “lower harm than alcohol” isn’t “no harm,” and people using cannabis for mental health deserve the full ledger.
How to Use This Information
A few practical principles, drawn from the evidence rather than the marketing:
- Talk to a prescribing clinician. If you’re considering cannabis for a diagnosed condition, that’s a conversation for a psychiatrist or primary care provider — ideally one informed about cannabinoid pharmacology. Most have seen patients self-medicate before; they won’t be shocked.
- Don’t substitute, augment. Where there is signal — schizophrenia adjunctive CBD, for example — it’s alongside conventional treatment, not instead. Stopping evidence-based medication because cannabis “feels like it’s working” is one of the fastest paths to relapse.
- CBD and THC are not the same. Most positive findings involve purified high-dose CBD or balanced CBD/THC products. Balance and Relax products are where therapeutic claims are most defensible; high-THC Uplift is where evidence is weakest and side-effect risk highest.
- Track your own response. The only way to know what works for you. Mood, sleep, anxiety, cognition — written down, not vibes.
- Watch for warning signs. Escalating dose, can’t enjoy things without cannabis, worse anxiety on tolerance breaks, flatter mood after sessions — pay attention.
- Practice harm reduction. Lower-THC products, limit frequency, keep tolerance breaks, avoid daily heavy use, never combine with alcohol or other depressants.
A Personal Note from Professor High
The cannabis world is full of confident claims that don’t survive contact with the evidence. Cannabis can be part of a good life. For some people — those with personal or family history of psychosis, those under 25, those with certain anxiety profiles — it can be actively harmful. For the broad middle, the evidence that it’s medicine in any rigorous sense is much weaker than the dispensary down the street would have you believe.
Mental Health Awareness Month is a good time to be honest about this. If you’re struggling, please talk to a real human professional — not your budtender, not a blog post.
Tracking your own response is the only way to know what works for YOU. The High IQ app lets you log strains, dosage, mood, and outcomes so you can see real patterns — not vibes. Free to start.
Sources
- Cooling S, Bonomo YA, Castle D, Hallinan CM. Randomised Controlled Trial Evidence on Medicinal Cannabis for Treatment of Mental Health and Substance Use Disorders: A Scoping Review. Clinical Drug Investigation 46(1):5–36 (2026). PMID 41343139. DOI · Full text (PMC)
- Wilson J, Dobson O, Langcake A, et al. The efficacy and safety of cannabinoids for mental disorders and SUDs: systematic review and meta-analysis. The Lancet Psychiatry (2026). DOI · Reuters coverage
- Black N, Stockings E, Campbell G, et al. Cannabinoids for mental disorders: systematic review and meta-analysis. The Lancet Psychiatry 6(12):995–1010 (2019). DOI
- McKee KA, et al. Cannabinoid products in adult psychiatric disorders: SR and meta-analysis of RCTs. J Psychiatric Research 140:267–281 (2021). DOI
- Whiting PF, et al. Cannabinoids for Medical Use: SR and Meta-analysis. JAMA 313(24):2456–2473 (2015). DOI
- de Bruijn SEM, et al. Differential effects of medicinal cannabis on mental health: SR. PMID 40186931. PubMed
- Cammà G, et al. Minor cannabinoids in psychiatric disorders: SR. Eur Neuropsychopharmacol 91:9–24 (2025). DOI
- ClinicalTrials.gov scoping review on cannabinoid-based products. Pharmaceutical Medicine 40:33–43 (2026). DOI
This article is educational, not medical advice. If you’re in crisis (US): call or text 988 (Suicide & Crisis Lifeline). International: findahelpline.com.