Cannabis users were about twice as likely to commit violence across 63 studies of 265,000 people

Sixty-three studies, 265,079 participants and one central number: in the general population, people who use cannabis had odds of committing a violent act about twice those of people who do not. The pooled odds ratio is 2.05, with a 95 percent confidence interval running from 1.74 to 2.41, and it comes from a systematic review and meta-analysis by a King’s College London team led by Giulia Trotta.

Trotta, who is at King’s College London, is the paper’s first and corresponding author, and Sir Robin Murray, a professor at King’s Institute of Psychiatry, Psychology and Neuroscience, is its senior author. It went online in Psychological Medicine on September 24, 2026.

Odds Ratios Across Four Kinds of Violence

The abstract on Cambridge Core splits the 63 studies by population and outcome. Among psychiatric patients, cannabis use carried an odds ratio of 2.49 for perpetrating violence. Where violence led to a criminal conviction, the ratios were larger: 3.75 in the general population and 4.21 among psychiatric patients. Cannabis users also had higher odds of being victims of violence, at 1.49, and the association was stronger in women and mixed-gender cohorts than in men.

An odds ratio is not the same as a doubling of risk. The two numbers sit close together only when the outcome is uncommon, and the review reports odds, so the “twice as likely” wording used by the authors and their university is a plain-English rendering of the 2.05 and not a measured doubling of absolute risk. The conviction figures rest on a smaller group of studies than the pooled estimate.

The Longitudinal Drop to 1.17

The study design matters more than the pooled figure suggests. Cross-sectional studies, which measure cannabis use and violence at the same moment, produced an odds ratio of 2.37. Longitudinal studies, which follow people over time and are better placed to say which came first, produced 1.17 (95% CI 1.07 to 1.28). The association survived, but it was much smaller, and most of the included studies, 46 of 64 analyses by the paper’s count, were cross-sectional.

The discussion section lists further limits. Violence was defined differently from study to study, covering physical, sexual, intimate and non-intimate forms, and cannabis exposure was captured by self-report, questionnaires and biological samples. Statistical heterogeneity was very high in most analyses, funnel plots were asymmetrical, and Egger’s tests flagged possible publication bias; adjusting for it with trim-and-fill reduced some estimates. Many of the included studies were rated only “fair” on quality.

Authors on Cause, Potency and Clinical Practice

King’s College London’s release, republished by SciTechDaily on October 2, is careful about causation. The review shows an association, and alcohol, other drug use and difficult social circumstances may account for part of it. Cannabis use also tends to co-occur with impulsivity and antisocial traits, a point the paper itself makes in discussing confounding.

Murray said the review shows that violence, both committing it and experiencing it, needs to be part of the conversation about cannabis. Marta Di Forti, a co-author at King’s and South London and Maudsley NHS Foundation Trust, said the team had been cautious for a long time about how far it could go in describing the link. The authors recommend that mental health services routinely ask patients about cannabis use when assessing violence risk, and that public health information mention the possible connection.

Potency is the open flank in the evidence. Most of the included research predates the spread of high-potency cannabis, and few studies recorded how much or how often people used. The authors suggest daily use of today’s stronger products could have larger effects, but the 63 studies could not test that. The review is funded by the NIHR Biomedical Research Centre at the Maudsley and the Medical Research Council (MR/T007818/1), per a summary of the paper dated September 30, and the Irish Health Research Board drugs library listing carries the same abstract figures.

Gizmodo, which covered the paper on October 5, quoted the authors as describing the link as likely reflecting “multiple interacting pathways rather than a single direct causal mechanism,” and listed possible routes the paper discusses, from impaired attention and inhibition to heightened paranoia with heavy use. Di Forti told the outlet that, as a clinician, she has seen heavy use contribute to violence, and described the association as “a factor” and not a proven cause. The Gizmodo account adds that most harms appear concentrated among chronic heavy users and not occasional ones, a reading the authors tie to the exposure-intensity gap in the underlying studies.

The search behind the review closed in November 2024, so any study published since then, including ones tracking modern high-THC products, sits outside the 265,079 people counted, and the authors’ own data cannot say whether weekly or daily use of today’s strongest products would push the 2.05 higher.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview