Cannabis and Violence Research: What Studies Show and Their Limits

Executive summary

The research record links cannabis use to higher rates of violence in some datasets, but it does not establish that cannabis causes violence, and population-level legalization data do not show a consistent statewide crime spike. A review summarized by ScienceDaily on September 30, 2026 carried the headline that cannabis users were twice as likely to commit violence. That figure is a relative association. It cannot show cause, and it says nothing on its own about absolute risk, dose, potency, or co-use of alcohol.

Three bodies of evidence frame the question:

• Individual-level studies compare users with non-users. They often find elevated violence among users, but they struggle to separate cannabis from alcohol, other drugs, mental illness, trauma, and antisocial traits.
• Population-level studies compare crime rates before and after state legalization. Studies of Colorado, Washington, and other states have generally not found a clear, attributable rise in violent crime.
• Clinical studies examine the link between high-potency THC, psychosis, and violence among a small vulnerable subgroup.

The topic matters commercially because it feeds ballot campaigns, potency-cap bills, insurance underwriting, and federal rescheduling. It matters clinically because the real signal, if there is one, probably sits in heavy, early-onset, high-potency, or co-substance use rather than in casual adult use. This hub sets out the history, the methods, the legal context, the state picture, and a framework operators and patients can use to read any new headline on the subject. Dated updates will be appended as new studies publish.

Why this matters

Violence findings shape policy, capital, and patient counseling at the same moment that federal rescheduling and state potency debates are live.

The stakes run across several groups.

• Regulators and legislators. A credible causal link between cannabis and violence would strengthen arguments for potency caps, purchase limits, and tighter licensing. A null result strengthens arguments for stability.
• Operators and investors. Multi-state operators (MSOs) already carry the burden of 280E tax treatment, limited banking, and price compression. A safety narrative that sticks adds reputational and underwriting risk on top of those costs.
• Patients and clinicians. Medical cannabis patients with psychotic disorders, bipolar disorder, or a history of aggression need accurate risk counseling, not slogans.
• Law enforcement and public health. Agencies need to know whether to expect more violent incidents near dispensaries or among heavy users, and whether to prioritize intoxication, withdrawal, or comorbid mental illness.

The scale is large. Dozens of states now permit adult-use or medical sales, and tens of millions of Americans report past-year use. Even a small absolute effect in a subgroup would translate into real numbers of incidents. A zero effect would also be meaningful, because it would remove one of the oldest arguments for prohibition.

Relative risk language is the main hazard. A "twice as likely" finding sounds alarming, but its meaning depends entirely on the baseline. As purely illustrative arithmetic (not data from any study), if 5 in 100 non-users committed a violent act in a given period, "twice as likely" would mean 10 in 100 among users. Whether that gap reflects cannabis, the people who choose cannabis, or something else is a separate question.

Background and history

The idea that cannabis drives violence is nearly a century old, and each generation of formal review has found the evidence thinner than the rhetoric.

1930s: the "reefer madness" era

The federal Marihuana Tax Act of 1937 passed amid press and government campaigns that portrayed cannabis as a cause of homicidal frenzy. Harry Anslinger, the first commissioner of the Federal Bureau of Narcotics, promoted anecdotes of violent crime attributed to the drug. Those anecdotes were not systematic research. The 1936 film Reefer Madness became shorthand for the genre.

1944: the La Guardia Committee

New York Mayor Fiorello La Guardia commissioned a study through the New York Academy of Medicine. The resulting 1944 report examined marijuana use in New York City and found that it did not lead to violent crime or to addiction in the way officials had asserted. Federal officials criticized it, but it remains the first major empirical challenge to the violence narrative.

1970: the Controlled Substances Act

Congress placed marijuana in Schedule I under 21 U.S.C. § 812, the schedule for drugs with a high potential for abuse and no accepted medical use. The Controlled Substances Act of 1970 also created the framework that the DEA and FDA still use for scheduling decisions. Violence was not the formal basis for the placement, but the surrounding political climate drew on the older narrative.

1972: the Shafer Commission

Congress created the National Commission on Marihuana and Drug Abuse, known as the Shafer Commission. Its March 1972 report, Marihuana: A Signal of Misunderstanding, concluded that cannabis was not a cause of violent crime and recommended decriminalizing personal possession. President Nixon rejected the recommendation. The Shafer finding is still cited by advocates, though it predates modern high-potency products.

1985: Goldstein's tripartite framework

Researcher Paul Goldstein proposed a model that remains the standard way to organize drug-violence research. It separates three pathways:

1. Psychopharmacological: the drug itself alters behavior toward aggression, as alcohol and stimulants can.
2. Economic-compulsive: users commit crime to fund a habit.
3. Systemic: violence arises from illegal market disputes, territory, and enforcement.

Cannabis scores weakly on the first two pathways in most analyses and strongly on the third, because prohibition-era markets generated violence through trafficking. That distinction matters for legalization, which removes the systemic pathway.

1990s-2000s: cohort studies and the psychosis question

Longitudinal cohorts, including birth cohorts in New Zealand and elsewhere, found associations between heavy adolescent cannabis use and later psychotic symptoms, antisocial behavior, and offending. Researchers debated direction of effect. Early antisocial traits predict both cannabis use and violence, which complicates any causal reading. Separately, the MacArthur Violence Risk Assessment Study tied violence among people with serious mental illness mainly to co-occurring substance misuse rather than to the illness alone.

2012-2014: Colorado and Washington open the natural experiment

Voters in Colorado (Amendment 64) and Washington (Initiative 502) approved adult-use legalization in November 2012. Colorado retail sales began January 1, 2014, and Washington's began in July 2014. Researchers gained a real-world test of whether legal markets raise violence.

2014-2019: the legalization crime literature

Morris and colleagues published a 2014 PLOS ONE analysis of state medical marijuana laws and found no increase in violent crime, with possible reductions in homicide and assault. Gavrilova, Kamada, and Zoutman reported in the Economic Journal (2019) that medical marijuana laws in states bordering Mexico were associated with reductions in violent crime, consistent with the systemic pathway. Dragone and colleagues examined Washington and Oregon and likewise did not report a violent crime surge. Findings varied by crime type and method, and local studies near dispensaries produced mixed results.

2017 onward: NASEM and the potency shift

The National Academies of Sciences, Engineering, and Medicine published The Health Effects of Cannabis and Cannabinoids in January 2017. It found substantial evidence linking cannabis use to an increased risk of developing schizophrenia and other psychoses, with the risk higher among the most frequent users. Di Forti and colleagues reported in Lancet Psychiatry (2019) that daily use of high-potency cannabis was associated with markedly higher odds of psychotic disorder in a multi-city European sample. Because untreated psychosis is itself a violence risk factor, these findings are the most plausible mechanism for an indirect cannabis-violence link.

2024-2026: rescheduling and renewed scrutiny

In May 2024 the DEA published an NPRM proposing to move marijuana to Schedule III, following a Health and Human Services recommendation. A formal hearing before an ALJ was scheduled for early 2025 and was then postponed. Federal executive action in late 2025 pushed the rescheduling process forward, and readers should check the current docket for the latest status. Against that backdrop, the September 30, 2026 ScienceDaily report on a review finding cannabis users twice as likely to commit violence re-opened the debate.

Year	Milestone	Violence finding or relevance
1937	Marihuana Tax Act	Passed amid violence-themed campaigns
1944	La Guardia Committee	No link to violent crime found
1970	Controlled Substances Act	Schedule I placement, 21 U.S.C. § 812
1972	Shafer Commission	Cannabis not a cause of violent crime
1985	Goldstein framework	Three pathways: pharmacological, economic, systemic
2012-2014	Colorado and Washington legalize	Natural experiment begins
2017	NASEM report	Psychosis association; frequent use raises risk
2019	Di Forti et al.	Daily high-potency use tied to psychosis
2024	DEA NPRM	Proposed Schedule III move
2026	ScienceDaily review report	Users "twice as likely" to commit violence

Key players

The debate is driven by federal health agencies, academic research groups, state regulators, industry trade groups, and prohibition-aligned advocacy organizations, each with a different evidentiary standard.

National Institute on Drug Abuse (NIDA)

NIDA funds most U.S. cannabis research and maintains public guidance on cannabis use disorder, psychosis risk, and adolescent exposure. Its materials emphasize that risks rise with early initiation, frequent use, and high potency.

DEA and FDA

The DEA holds scheduling authority under the Controlled Substances Act, while the FDA and HHS supply the scientific and medical evaluation. Their review of abuse potential and public safety shapes any final rescheduling rule. Violence is not a formal scheduling factor, but safety evidence enters the record through public comment.

National Academies of Sciences, Engineering, and Medicine

NASEM's 2017 committee remains the most-cited consensus review. It graded evidence by strength, and it is the baseline against which new reviews are compared.

Academic research groups

Psychiatric epidemiology teams, including the group around Marta Di Forti on cannabis and psychosis, drive the potency research. Economists such as Gavrilova, Kamada, and Zoutman, and criminologists such as Morris and colleagues, drive the legalization-and-crime literature. Their methods differ, which is why their conclusions can look inconsistent.

State regulators