Does Cannabis Cause Violence? What a New Review Can Tell Us

Does cannabis use increase the likelihood of violence? A new review reports an association, but answering that question requires more than repeating its headline numbers. The findings change substantially depending on who was studied and how researchers measured cannabis use and violent behaviour.

A systematic review and meta-analysis by Giulia Trotta and colleagues, published in Psychological Medicine on September 24, 2026, examined 63 studies involving 265,079 participants.¹ Cannabis use was associated with higher odds of both committing violence and experiencing it. However, the review cannot establish that cannabis itself caused those outcomes.

For consumers, clinicians, and policymakers, the useful question is what explains the association. Drug effects, mental health, other substance use, and social environments may all contribute. Understanding their relative importance determines which interventions could actually improve safety.

What the Cannabis and Violence Review Found

The researchers searched three scientific databases for studies published through November 2024. They examined violence perpetration and victimization separately, while distinguishing general-population samples from psychiatric populations.

Across general-population studies, cannabis users had approximately twice the odds of perpetrating violence compared with non-users. The pooled odds ratio was 2.05. Among psychiatric populations, the corresponding estimate was 2.49. Cannabis use was also associated with higher odds of becoming a victim of violence, with an overall estimate of 1.49.

These are averages across studies using different definitions of cannabis exposure and violence. Some measured recent use, while others measured lifetime use. Outcomes included physical, sexual, and psychological violence, intimate partner violence, and criminal convictions. The combined results therefore describe a broad relationship rather than one clearly defined situation.

Higher Odds Do Not Describe an Individual’s Probability

An odds ratio compares the odds of an outcome between groups. It does not tell readers how many cannabis users committed violence, nor does an odds ratio of 2.05 necessarily mean their probability was doubled.

Absolute risk depends on how common the outcome was in the comparison group. The same odds ratio can accompany a small or substantial difference in actual probability. Without that baseline, the pooled estimate cannot become a personal forecast. It also cannot support the claim that most cannabis users are violent.

Why Study Design Changes the Story

The most revealing contrast appears within the general-population perpetration results. Cross-sectional studies produced an odds ratio of 2.37, while longitudinal studies produced an estimate of 1.17.

Cross-sectional studies typically measure exposure and outcomes at one point or over overlapping periods. They can identify whether cannabis use and violence occur together, but often cannot establish which came first. Longitudinal studies follow participants over time, making the sequence easier to assess.

The smaller longitudinal association remained statistically significant. Nevertheless, the difference matters: the largest headline estimate does not describe the association found when general-population participants were followed over time.

That pattern was not universal. Psychiatric-population longitudinal studies produced a larger estimate of 3.71, although this subgroup contained only four studies. General-population results therefore should not be substituted for clinical findings, or vice versa.

In a September 29 expert assessment of the review, Cardiff University public health researcher Simon Moore highlighted the importance of social context. He argued that the analysis could not distinguish cannabis effects from the environments in which people used it, including peer groups, neighbourhood conditions, and exposure to illegal markets.

The implication is that prevention needs a clearer explanation of the relationship. If substance effects and environmental exposure contribute differently across groups, one intervention may not address both.

Association Does Not Identify the Cause

The review found similar general-population estimates when comparing studies reporting adjusted and unadjusted odds ratios. Adjusted analyses attempt to account for other variables that could influence the relationship.

However, adjustment only addresses factors researchers measured and included. Similar adjusted results do not demonstrate that every important influence was removed.

The authors acknowledge that most included studies did not comprehensively account for accompanying alcohol or other drug use. Differences in impulsivity, antisocial traits, social circumstances, and mental health could also affect both cannabis consumption and violence.

Consider two people reporting cannabis use. One consumes occasionally at home. Another uses several substances within a peer group where violence is already common. Classifying both simply as cannabis users can obscure the very circumstances needed to explain their outcomes.

This does not eliminate the possibility of a causal contribution from cannabis. It shows why an observed relationship can contain several pathways, and why pooling studies cannot recover details those studies never collected.

What the Findings Mean for Mental Health and Victimization

The higher pooled estimate in psychiatric populations makes clinical assessment particularly relevant. The authors recommend routinely assessing cannabis use among psychiatric patients.

They discuss possible pathways involving psychotic symptoms, impaired judgement, emotional regulation, and withdrawal-related irritability. These mechanisms could help explain some associations, but the meta-analysis did not demonstrate which pathway produced a particular violent event.

A psychiatric diagnosis or cannabis-use history alone is therefore an inadequate individual safety assessment. The practical value lies in examining symptoms, consumption patterns, other substances, and circumstances together.

Victimization Deserves Equal Attention

The association with becoming a victim broadens the public health question. General-population longitudinal victimization studies produced an odds ratio of 1.05, substantially smaller than the cross-sectional estimate of 2.20.

Possible explanations include shared environmental exposure, vulnerability, or behavioural effects. The findings do not establish that cannabis use caused someone’s victimization, and they do not transfer responsibility away from perpetrators.

For services supporting people exposed to violence, cannabis use may be one relevant part of a wider assessment. Treating it as the whole explanation could overlook unsafe relationships or other conditions that require direct intervention.

The Missing Product Data Matter

Most included studies lacked sufficient information about consumption frequency, dose, duration, potency, or THC-to-CBD composition. The researchers consequently could not conduct meaningful dose-response or potency subgroup analyses.

That prevents several tempting conclusions. The review cannot establish whether occasional use differs from daily use, whether concentrates differ from lower-potency flower, or whether CBD changes the association.

More generally, MyCannabis has examined why cannabis research still lags behind the science, including the challenges created by inconsistent products and research infrastructure. This review illustrates the corresponding problem in safety research: a broad exposure label can conceal important differences.

Better measurement also requires dependable product information. MyCannabis reported on a recent Massachusetts cannabis potency audit. The regulator’s August 2026 published audit results identified 13 of 63 flower products outside its acceptable labelling range. That audit does not explain violence, but it shows why researchers need reliable exposure measurements before comparing product-specific outcomes.

What Cannabis Policy Can Learn From the Review

The findings support further investigation of violence within cannabis-related harm assessment. They do not demonstrate that legalization increases violence. Comparing users with non-users answers a different question from comparing outcomes before and after a policy change.

A convincing policy evaluation would need to examine consumption patterns, product availability, enforcement, illegal-market exposure, and other changing conditions. Without those distinctions, the same association could be used to support competing arguments without resolving either.

Future research would be more useful if it consistently recorded:

• Frequency, dose, potency, and product composition.
• Alcohol and other substance use.
• Mental health and prior violence.
• Social setting and the timing of events.

The strongest takeaway is that the association warrants attention, while its explanation remains incomplete. Responsible coverage should preserve both points. Better evidence must identify which people, patterns, and circumstances account for harm so that prevention can target them effectively.

References:

¹ Trotta, G., Rodriguez, V., Marino, P., Gurmesa, M., Spinazzola, E., Li, Z., Alameda, L., Di Forti, M., Murray, R., & Vassos, E. (2026). Cannabis use is associated with increased risk of violence: A systematic review and meta-analysis. Psychological Medicine, 56, Article e296. https://doi.org/10.1017/S0033291726105856