Cannabinoids are often discussed as if they are one treatment for one broad category called mental health. The new review in The Lancet Psychiatry shows why that framing is too simple. The answer changed with the diagnosis, outcome, cannabinoid formulation, and trial design.
The review included 54 randomized controlled trials and 2,477 participants. Signals appeared for cannabis use disorder, insomnia, tic or Tourette syndrome, and autistic traits. Most evidence was low certainty, 44% of trials had a high risk of bias, and participants receiving cannabinoids had higher odds of any adverse event.
What the Researchers Asked
The team searched five research databases for peer-reviewed randomized controlled trials published from January 1, 1980, through May 13, 2025. To qualify, a cannabinoid had to be the primary treatment for a mental disorder or substance use disorder. The main outcomes were remission or a reduction in symptoms.
That inclusion rule matters. The review was not asking whether a person with chronic pain also slept or felt better. It focused on trials designed to treat the psychiatric or substance use condition itself.
The included studies enrolled 2,477 people with a median age of about 33. Men made up 69% of participants, and the publications did not provide usable ethnicity data. The review was registered in advance with PROSPERO, used two independent reviewers, assessed risk of bias with the Cochrane tool, and rated evidence certainty with GRADE.
Where the Review Found Signals
The positive findings were narrow and condition-specific. They should not be translated into a claim that cannabis generally improves mental health.
Possible benefit
Cannabis use disorder
A combined CBD and THC treatment produced small reductions in withdrawal symptoms and reduced weekly cannabis use by about one gram compared with placebo.
Possible benefit
Insomnia
Across cannabinoid types, people with insomnia recorded more sleep time using both electronic devices and sleep diaries.
Possible benefit
Tics and Tourette syndrome
Pooled results showed lower tic severity than placebo. The estimate was larger than several other positive signals, but the evidence base remained limited.
Possible benefit
Autistic traits
The review found a modest reduction in measured autistic traits. This does not establish that cannabinoids broadly improve daily function or quality of life for autistic people.
Where the Evidence Did Not Show Benefit
The pooled analyses found no statistically significant benefit for anxiety, anorexia nervosa, psychotic disorders, post-traumatic stress disorder, or opioid use disorder. For cocaine use disorder, cannabinoid treatment was associated with more cocaine craving, not less.
There were too few data to pool results for ADHD, bipolar disorder, obsessive-compulsive disorder, or tobacco use disorder. The researchers found no eligible randomized controlled trial evidence for depression.
No eligible trial evidence for depression does not mean a trial proved cannabinoids ineffective for depression. It means the review did not find randomized evidence capable of answering the question.
Side Effects Were More Common
Participants assigned to cannabinoids had 1.75 times the odds of experiencing any adverse event compared with control groups. The authors calculated a number needed to harm of seven. In practical terms, about one additional person experienced an adverse event for every seven people treated instead of given the comparison intervention.
The review did not find higher odds of serious adverse events or withdrawal from a study. That is reassuring, but it does not make the overall safety finding disappear. In the insomnia trials, commonly reported adverse events included dry mouth, nausea, diarrhea, and dizziness. The exact adverse-event profile also depends on the condition, product, dose, route, and individual.
Why Confidence Remains Low
Twenty-four of the 54 trials were rated at high risk of bias, and certainty was low for most outcomes. The studies also grouped together a wide range of products, including different THC and CBD ratios, doses, routes, and treatment periods. A standardized medicine used in a trial is not interchangeable with retail flower, a gummy, or an unverified CBD tincture.
The participant pool also limits how confidently the findings can be generalized. More than two-thirds of participants were male, ethnicity data were unavailable, and many condition-specific conclusions relied on small numbers of trials. A meta-analysis can make a scattered evidence base easier to read, but it cannot repair weak or highly varied underlying studies.
The work was funded by Australia’s National Health and Medical Research Council. The paper reports consultation fees from the World Health Organization for two authors, expert-testimony payments related to cannabis risk for one author, and government advisory or evidence-review roles for another. The remaining authors declared no competing interests.
What This Means for Patients and Clinicians
The strongest conclusion is not that cannabinoids never help. It is that broad psychiatric claims move faster than the randomized evidence. The signals for insomnia, cannabis use disorder, tics, and autism deserve better trials with standardized products, longer follow-up, outcomes that matter in daily life, and more representative participants.
For now, a condition-specific conversation is more useful than asking whether cannabis is good or bad for mental health. Patients should know the exact product and dose being considered, what outcome would count as meaningful improvement, what side effects will be monitored, and how the plan fits with existing treatment. Do not stop or replace prescribed mental health treatment based on a review headline.
The Bottom Line
This is one of the broadest recent summaries of randomized cannabinoid evidence for mental health and substance use disorders. It found several legitimate research signals, but not a general case for routine treatment. The authors concluded that routine cannabinoid use for these disorders is rarely justified by the current evidence. That conclusion reflects limited confidence, not a refusal to study possible benefits.
Clear answers
Frequently Asked Questions
Does this review show that cannabis treats mental illness?
No. The review found a few condition-specific signals, but evidence for most outcomes was low certainty. The authors concluded that routine cannabinoid treatment for mental disorders and substance use disorders is rarely justified by the current evidence.
Which conditions showed a possible benefit?
The pooled results found signals for cannabis use disorder, insomnia, tic or Tourette syndrome, and autistic traits. These findings came from different cannabinoid products and generally low-certainty evidence, so they should not be applied to every cannabis or CBD product.
What did the review find for anxiety, PTSD, and depression?
The meta-analysis found no statistically significant benefit for anxiety or post-traumatic stress disorder. It found no eligible randomized controlled trial evidence for depression, which is different from proving that cannabinoids do not work for depression.
Were side effects more common?
Yes. The pooled odds of experiencing any adverse event were higher with cannabinoid treatment, with a reported number needed to harm of seven. The review did not find higher odds of serious adverse events or withdrawal from a study.
Can a retail cannabis product be expected to reproduce these results?
No. The trials used different formulations, cannabinoid ratios, doses, routes, and treatment periods. A retail flower, gummy, tincture, purified CBD medicine, and standardized THC-CBD product are not interchangeable.