The Largest Review of Cannabis and Mental Health Finds Almost No Evidence It Works

Two sweeping new analyses — one spanning 54 clinical trials over 45 years — conclude that the science does not support cannabis as a treatment for anxiety, depression, or PTSD, even as millions of patients are prescribed it for those very conditions.

The Studies

On March 17, 2026, researchers at the University of Sydney's Matilda Centre for Research in Mental Health and Substance Use published the largest-ever systematic review and meta-analysis of cannabinoids for mental health and substance use disorders in The Lancet Psychiatry [1]. Led by Dr. Jack Wilson, the team analyzed 54 randomized controlled trials published between 1980 and 2025, encompassing data from 2,477 participants across a wide range of psychiatric conditions and cannabinoid formulations [2].

The same week, a companion review in JAMA Internal Medicine, co-authored by Dr. Deepak Cyril D'Souza of Yale University's Department of Psychiatry and the VA Connecticut Healthcare System, took a broader lens — incorporating observational studies and other research designs alongside clinical trials — and arrived at substantially the same conclusion [3].

The verdict from both papers is blunt: there is no good evidence that any form of cannabis is effective for treating the mental health conditions for which it is most commonly prescribed.

What the Data Actually Shows

The Lancet Psychiatry analysis applied strict inclusion criteria, accepting only randomized controlled trials — the gold standard for evaluating medical treatments. The results were striking in their absence of positive findings [1][2]:

• Anxiety disorders: No evidence of benefit
• Depression: Not a single clinical trial met the review's criteria for inclusion
• PTSD: No evidence of benefit
• Psychotic disorders (including schizophrenia): No evidence of benefit
• Bipolar disorder: No evidence of benefit; cannabis may worsen manic episodes
• Anorexia nervosa: No evidence of benefit
• OCD: No evidence of benefit

A handful of conditions showed tentative positive signals — insomnia, autism spectrum disorder, and tics or Tourette's syndrome — but the authors rated even this evidence as "low quality" due to small sample sizes and methodological limitations [2][4].

One finding carried a direct warning: when medicinal cannabis was used to treat cocaine-use disorder, it increased cravings, suggesting it could actively worsen dependence in some populations [4].

"In the absence of evidence at this time, the routine use of medical cannabis products really should be rarely justified" for mental health disorders, Wilson told NPR [1].

Source: GDELT Project

Data as of Mar 18, 2026CSV

A 45-Year Research Failure

Perhaps the most damning finding is not what the studies show, but what they don't. After 45 years of published research, only 54 trials met even basic quality standards — and for depression, one of the most common reasons patients seek medical cannabis, there was not a single qualifying trial [1].

The numbers are startling in their inadequacy. For anxiety and PTSD combined, just 112 participants were studied across eight small trials, five of which enrolled 13 or fewer people. Only two evaluated patients for a clinically meaningful period of one month [5].

"It's embarrassing how little we have done in terms of data collection, given how widely this is available as a therapeutic," said Ryan Vandrey, a professor at Johns Hopkins University who studies cannabis pharmacology [1].

Tory Spindle, also at Johns Hopkins, put it in broader pharmaceutical context: "For a long time, people have been using cannabis to cope with mental illness. And what is really lacking is controlled, rigorous evidence, which would be the same standards you would want for any medication" [5].

The research gap is not accidental. Cannabis remains federally classified as a Schedule I substance in the United States, which has historically created bureaucratic and legal barriers to conducting clinical trials. But even as 40 states and Washington, D.C. moved to legalize medical marijuana — and as the cannabis industry grew into a multi-billion-dollar market — the parallel investment in rigorous clinical research never materialized [1][6].

The Policy-Evidence Disconnect

The gap between what the science shows and what the law permits is vast. Forty states plus Washington, D.C. now allow medical cannabis, and at least 12 states explicitly list PTSD as a qualifying condition [6][7]. All but two state medical cannabis programs — Alaska and South Dakota — permit cannabis for PTSD [7]. Anxiety, depression, and chronic pain are among the most frequently cited reasons patients obtain medical marijuana cards across the country.

In Australia, where the Lancet study originated, the disconnect is equally stark. Over one million medicinal cannabis prescription approvals have been issued in just four years, with sales tripling over that period and more than 700,000 Australians using medicinal cannabis for over 250 different conditions [4].

"These two papers clearly show there isn't any evidence to recommend the use of cannabis or cannabis derivatives to treat mental health," said D'Souza of Yale [3]. "We all just want people to be able to access medicines that are both effective and safe for their conditions. Unfortunately, for a lot of people using medical cannabis, I don't think that is the case," Wilson added [4].

The regulatory frameworks that enabled this expansion were, in most cases, built on political momentum and patient advocacy rather than clinical evidence. States adopted qualifying conditions piecemeal, often through legislative negotiation rather than systematic evidence review. A 2024 study comparing state qualifying conditions to the National Academies of Sciences' 2017 evidence report found significant misalignment — states routinely approved conditions for which the NAS found insufficient or no evidence of benefit [8].

Risks That Are Better Documented Than Benefits

While the evidence for cannabis as a mental health treatment is thin, the evidence for cannabis as a mental health risk factor is considerably stronger. Both reviews warn of "substantial risks" for vulnerable populations [1][3]:

• Adolescents and young adults: A well-documented link exists between youth cannabis use and increased risk of psychosis, with higher-potency products carrying greater danger
• People with bipolar disorder: Cannabis can worsen manic symptoms
• People with psychotic spectrum disorders: Cannabis can exacerbate psychotic symptoms in conditions such as schizophrenia
• Substance use vulnerability: An estimated 30 percent of people who use cannabis develop cannabis use disorder, a recognized mental health condition characterized by problematic patterns of use leading to significant impairment [9]

Wilson noted that routine medicinal cannabis use "could be doing more harm than good by worsening mental health outcomes, for example a greater risk of psychotic symptoms and developing cannabis use disorder, and delaying the use of more effective treatments" [4].

This last point — the opportunity cost of cannabis use — may be the most underappreciated risk. When patients turn to cannabis for anxiety or depression instead of pursuing evidence-based treatments like cognitive behavioral therapy or established psychiatric medications, they may be losing valuable time in managing their conditions.

What Cannabis Can Treat

The picture is not uniformly negative. The Lancet review did identify conditions where cannabinoids have established or emerging evidence of benefit — they simply aren't the mental health conditions driving most prescriptions [2][4]:

• Epilepsy: Cannabidiol (CBD) has demonstrated seizure reduction, and the FDA-approved drug Epidiolex is based on this evidence
• Multiple sclerosis spasticity: Modest evidence supports cannabinoid use
• Certain chronic pain conditions: Some evidence, though contested
• Cannabis use disorder itself: Cannabinoids combined with psychological therapy showed potential benefit — an ironic finding given the broader context

The distinction between cannabidiol (CBD) and tetrahydrocannabinol (THC) matters here. CBD, which does not produce the psychoactive "high" associated with cannabis, has shown promise through its action on serotonin receptors. But much of what is prescribed and consumed as medical marijuana contains THC, and the two compounds have markedly different risk profiles [5].

What Comes Next

The studies' authors and independent experts agree on one point: the answer is not to dismiss cannabis research but to demand more of it. The current state of evidence reflects decades of regulatory obstruction, commercial disinterest in funding rigorous trials (since cannabis cannot be patented like a pharmaceutical), and a political landscape that moved faster than the science.

Wilson and his colleagues called for strengthened regulation of medicinal cannabis prescribing practices and evidence-based clinical decision-making to minimize harm from ineffective products [4]. D'Souza emphasized that clinicians should communicate the current evidence honestly to patients who ask about cannabis for mental health [3].

For the roughly 15 percent of American adults who are active marijuana users — and the subset who use it specifically to manage anxiety, depression, or PTSD — the findings do not necessarily mean cannabis provides no subjective relief [9]. Placebo effects are real, and individual responses vary. But the studies make clear that, by the standards applied to every other medication on the market, the evidence simply is not there.

The broader question is whether the legal and commercial infrastructure built around medical cannabis will adapt to the science, or whether the science will continue to trail behind policy and profit. As Vandrey put it: after 45 years, the embarrassment isn't that the evidence is weak — it's that so little effort was made to find it [1].