Peer-Reviewed Study Finds No Traffic Fatality Increase After Legalization

Study Design and Methodology

Researchers analyzed 19 years of traffic fatality data from states that legalized adult-use cannabis, comparing crash rates before and after implementation using a difference-in-differences model with imputation controls. The study controlled for variables including population density, alcohol consumption rates, seat-belt laws, and economic conditions. Data spanned 2005 through 2024, covering early-legalizing states like Colorado and Washington as well as more recent adopters.

The imputation-based approach allowed researchers to account for missing or incomplete crash data in certain jurisdictions. Incomplete datasets have plagued earlier traffic-safety studies, leading to contradictory findings that muddied the policy debate. This matters.

Core Findings: No Fatality Spike Detected

The study found no statistically significant change in overall traffic fatality rates in states that legalized recreational cannabis compared to control states. When researchers isolated the post-legalization period, fatality trends tracked closely with national baseline rates adjusted for population growth and vehicle miles traveled.

This holds even when broken out by subgroups: nighttime crashes, single-vehicle accidents, and crashes involving drivers under 25 showed no meaningful divergence from pre-legalization trends. The null finding is the headline. It's what two decades of data failed to produce.

What the Study Doesn't Claim

The authors didn't conclude that cannabis has no impairing effect on driving—only that legalization itself didn't measurably increase crash fatalities at the population level. Individual impairment is well-documented in laboratory settings and roadside testing. The disconnect is scale: legalization changes access, not pharmacology, and the fatality data suggest offsetting behaviors or enforcement adaptations may be at work.

One hypothesis: legal markets displace black-market consumption that was already happening, rather than creating net-new impaired drivers. The study doesn't test that mechanism directly.

Implications for State Policy Debates

The findings arrive as several states—including Ohio and Pennsylvania—work through post-legalization rulemaking around impaired-driving thresholds and roadside testing protocols. Prohibitionist groups have leaned heavily on traffic-safety arguments to oppose legalization or demand restrictive per-se THC limits. This study undercuts that framing with longitudinal data.

Ohio's adult-use market launched in August 2024. Pennsylvania's legislature is debating a legalization framework for 2027. Both states have cited traffic safety as a primary concern in public hearings. The Cureus study gives legalization advocates empirical cover, though it won't settle the impairment-testing debate—THC nanogram limits remain scientifically contentious.

Contrast With Earlier Meta-Analyses

A 2022 meta-analysis published in Accident Analysis & Prevention found mixed results, with some studies showing modest fatality increases and others showing decreases or no change. The inconsistency stemmed from methodological differences: short observation windows, failure to control for alcohol policy changes, and incomplete crash datasets. The Cureus study's 19-year span and imputation controls address those gaps.

Earlier literature also struggled with timing—many studies captured only the first 2-3 years post-legalization, when regulatory frameworks were still stabilizing. By extending the window to 2024, the new study captures mature markets in Colorado, Washington, Oregon, and California.

Researcher Attribution and Limitations

The study was conducted by a multi-institutional research team and underwent peer review before publication in The Cureus Journal of Medical Science, a PubMed-indexed open-access journal. The authors disclosed no conflicts of interest and noted that the study's observational design can't establish causality—only association. Unmeasured confounders, such as changes in ride-sharing availability or autonomous-vehicle adoption, could influence results.

The paper also flags a data limitation: THC blood-concentration data from crash scenes remain inconsistent across states, making it impossible to disaggregate cannabis-involved crashes from the broader fatality pool with precision. That's a structural gap in U.S. crash reporting. Not a flaw in this study's design.

What to Watch: Federal Rescheduling and Interstate Data Sharing

If the DEA finalizes cannabis rescheduling to Schedule III in 2026, federal funding for state-level crash-data standardization could improve, enabling more granular future studies. For now, the Cureus findings are the most comprehensive longitudinal read available. Operators and policymakers should treat the null result as signal, not noise—it's what 19 years of real-world data actually show.

Next variable to watch: how states with newly operational markets—like Ohio and Minnesota—trend over the next 3-5 years. If the pattern holds, the traffic-safety objection to legalization loses its empirical foundation. For full background on this story, see the CannIntel topic hub on Cannabis and Traffic Safety.