A systematic review and meta-analysis of 9 observational studies found cannabis use was associated with a 1.92 times increased risk of motor vehicle collisions, with fatal crashes showing a 2.10 times increase.
Read this if you want to know the actual crash risk data for driving after cannabis use.
1.92x increased crash risk with cannabis use
The Backstory
The question had been circulating in emergency rooms, traffic safety agencies, and state legislatures for years: how dangerous is it to drive after using cannabis?
Everyone had opinions. Cannabis advocates pointed out that stoned drivers tend to slow down and increase following distance — the opposite of drunk drivers. Law enforcement argued that any impairment was unacceptable behind the wheel. And researchers kept producing studies with wildly different conclusions, ranging from "minimal risk" to "significantly elevated risk."
What was missing was a rigorous synthesis of all the available evidence. In 2012, Mark Asbridge and colleagues at Dalhousie University in Nova Scotia set out to provide one.
The Study
Asbridge, Hayden, and Cartwright conducted a systematic review and meta-analysis — the gold standard for synthesizing research evidence — published in the BMJ, one of the world's most prestigious medical journals.
Study Design
The Definitive Meta-Analysis of Cannabis and Crash Risk
19
Databases searched
with no year or language restrictions
9
Studies included
observational studies with appropriate control groups
Toxicology
Cannabis measured by
blood THC or self-report of recent use
Random effects
Statistical model
accounting for heterogeneity between studies
Asbridge et al. (2012), BMJ 344:e536
The search was exhaustive: 19 electronic databases with no restrictions on year or language, plus manual reference searches, unpublished study searches, and the research team's personal libraries. They included only observational studies of actual motor vehicle collisions — not simulator studies or experimental driving tests — that had appropriate control groups and measured recent cannabis use through blood toxicology or self-report.
Nine studies met the inclusion criteria. Each was assessed for risk of bias using the Newcastle-Ottawa scale, and results were combined using random effects meta-analysis.
The Finding
1.92×
the odds ratio for motor vehicle collision while driving under the influence of cannabis compared to unimpaired driving (95% CI: 1.35-2.73, p = 0.0003). Cannabis nearly doubles crash risk.
For context, alcohol at the legal limit (0.08 BAC) increases crash risk approximately 7-14 fold. Cannabis impairment is real but operates in a fundamentally different magnitude than alcohol impairment.
Asbridge et al. (2012), BMJ
The result was consistent and statistically robust: acute cannabis use roughly doubled the risk of a motor vehicle collision. But the subgroup analyses revealed important nuances.
Subgroup Findings
Not All Crashes Are Equal
Fatal crashes
OR 2.10 (95% CI 1.31-3.36) — statistically significant. Cannabis-involved crashes are disproportionately fatal.
Non-fatal crashes
OR 1.74 (95% CI 0.88-3.46) — NOT statistically significant. The risk increase for non-fatal crashes was less clear.
Case-control studies
OR 2.79 — higher estimates, possibly reflecting better exposure measurement
Culpability studies
OR 1.65 — lower estimates, closer to the minimum true risk
Heterogeneity
I² = 81% — substantial variation between studies, meaning the true risk may vary significantly depending on context
Asbridge et al. (2012), BMJ
The high heterogeneity (I² = 81%) is important. It means the nine studies produced substantially different estimates, and the pooled 1.92 figure is an average across studies with real methodological differences. Some studies used blood THC to confirm recent use; others relied on self-report. Some looked at fatal crashes; others included fender-benders. The true risk likely varies with dose, tolerance, time since use, and whether alcohol is also involved.
Cannabis vs. Alcohol: A Fundamentally Different Impairment
The comparison to alcohol is the single most important piece of context for understanding cannabis impairment.
The impairment profiles are qualitatively different as well. Alcohol impairs judgment, risk perception, and inhibition — drunk drivers drive faster, tailgate more, and take more risks. Cannabis impairs reaction time, attention, and tracking — but cannabis users tend to be aware of their impairment and compensate by driving slower and increasing following distance. This compensatory behavior partially offsets the psychomotor impairment, which is why the crash risk increase is much smaller than for alcohol.
This does not make cannabis-impaired driving safe. A nearly doubled crash risk is meaningful, particularly for fatal collisions. But it does mean that equating cannabis impairment with alcohol impairment — as some DUI laws effectively do — misrepresents the relative risk.
The THC Blood Level Problem
One of the most consequential policy implications of this research is the failure of THC blood levels to correlate with actual impairment.
Myth vs. Reality
A THC blood level of 5 ng/mL is equivalent to a blood alcohol concentration of 0.08 — both indicate impairment
THC pharmacokinetics are fundamentally different from alcohol. Peak impairment occurs roughly 90 minutes after smoking, by which time blood THC has already declined more than 80% from its peak. A regular user may have 5 ng/mL with no impairment at all, while an occasional user may be significantly impaired at 2 ng/mL. The research director of the Washington Traffic Safety Commission has called the 5 ng/mL threshold 'completely arbitrary' and 'not backed by scientific theories.' AAA concluded that per se THC limits are 'unsupported by science.'
The Evidence
AAA Foundation for Traffic Safety; NHTSA Report to Congress (2017)
This creates an impossible policy dilemma. Five states have adopted per se limits (typically 5 ng/mL), but these limits miss an estimated 70% of cannabis-impaired drivers while potentially criminalizing sober regular users who happen to have residual THC in their blood. The Asbridge meta-analysis provided the crash risk data that policymakers needed, but the detection technology has not kept pace with the science.
What the Study Could Not Tell Us
Limitations
What We Still Don't Know
Dose-response
The meta-analysis could not determine how crash risk changes with dose. One hit vs. a full session may produce very different impairment levels.
Tolerance effects
Regular users develop tolerance to many of THC's psychomotor effects. Whether daily users drive as impaired as occasional users remains unclear.
Time course
The studies measured 'recent use' differently — some within hours, some only testing positive for THC. The time window of maximal driving risk is not precisely defined.
THC vs. THC + CBD
CBD may attenuate some of THC's impairing effects. Modern products with varying THC:CBD ratios may produce different driving impairment profiles.
Cannabis + other substances
The combined risk of cannabis with alcohol is multiplicative (~15-20x), not additive. Most real-world cannabis-impaired driving involves polysubstance use.
Asbridge et al. (2012), BMJ
The Policy Impact
The Asbridge meta-analysis was cited over 100 times in the two years following publication and directly influenced drug-impaired driving policies in Canada and beyond. It informed changes to graduated driver licensing laws in Nova Scotia, the Canadian Centre on Substance Abuse's anti-drug driving campaign, and ultimately contributed to the evidence base for Canada's federal cannabis impairment laws when legalization occurred in 2018.
The study's conclusion was careful and policy-oriented: "This information could be used as the basis for campaigns against drug impaired driving, developing regional or national policies to control acute drug use while driving, and raising public awareness." It did not advocate for specific legal limits — a wise choice given the THC blood level measurement problems that continue to vex legislators.
The Researcher
Mark Asbridge is a professor of community health and epidemiology at Dalhousie University in Halifax, Nova Scotia. His research program focuses on injury epidemiology, substance use, and traffic safety — an intersection that has made him one of the most influential voices in the cannabis-impaired driving evidence base. The Dalhousie research context — a Canadian university in a country that would legalize cannabis six years after this publication — adds particular relevance to this work.
How dangerous is it to drive after using cannabis?
Cannabis roughly doubles the risk of a motor vehicle collision based on this meta-analysis. For fatal crashes, the risk is about 2.1 times higher. This is a meaningful increase but dramatically lower than alcohol impairment, which increases risk 7-14 fold at the legal limit. The combination of cannabis and alcohol is far more dangerous than either alone.
How long should I wait after using cannabis before driving?
There is no scientifically validated safe waiting period. Impairment varies dramatically by individual, dose, tolerance, and consumption method. Peak impairment from smoking typically occurs around 60-90 minutes and can last 3-4 hours. Edibles can impair for 6-8 hours or longer. If you feel any cognitive effects, you should not drive. Our driving impairment guide covers this in more detail.
Can I get a DUI for cannabis?
Yes, in every US state and Canadian province. The specifics vary: some jurisdictions use per se THC blood limits, others require demonstrated impairment, and some use both. The legal landscape is complex and the measurement science is deeply problematic. The safest approach is simple: if you have used cannabis, do not drive.
Is cannabis-impaired driving as dangerous as drunk driving?
No. The data consistently shows that cannabis impairment produces a much smaller increase in crash risk than alcohol. Cannabis roughly doubles risk; alcohol at the legal limit increases it 7-14 fold. However, cannabis impairment is still real and still dangerous — especially for fatal crashes, and especially when cannabis and alcohol are combined.
Acute cannabis consumption and motor vehicle collision risk: systematic review of observational studies and meta-analysis
Asbridge M, Hayden JA, Cartwright JL (2012) · BMJ
Related Research
Key studies in this area
Triggering Myocardial Infarction by Marijuana
Mittleman et al. (2001)
Cannabis and Driving: The Original Evidence
Asbridge et al. (2012)
What the researchers found
Researchers searched 19 databases without language or year restrictions and identified 9 observational studies meeting inclusion criteria. Cannabis use while driving was associated with a significantly increased collision risk (odds ratio 1.92, 95% CI 1.35-2.73, p=0.0003).
Risk estimates were higher in studies of fatal collisions (OR 2.10) than non-fatal ones (OR 1.74, not statistically significant). Case-control studies showed higher estimates (OR 2.79) than culpability studies (OR 1.65).
Heterogeneity among studies was high (I-squared 81%), indicating substantial variation in results across different study designs and populations.
Why it matters
This was one of the most comprehensive meta-analyses of cannabis-impaired driving risk. The near-doubling of crash risk provided a clear evidence base for policy on drug-impaired driving.
The numbers in context
Overall OR: 1.92 (95% CI 1.35-2.73). Fatal collisions: OR 2.10 (1.31-3.36). Non-fatal: OR 1.74 (0.88-3.46). Case-control studies: OR 2.79. Culpability studies: OR 1.65. I-squared: 81%.
How the study worked
Systematic review and meta-analysis. Searched 19 electronic databases with no year or language restrictions, plus manual searches and unpublished studies. Included observational studies with appropriate control groups measuring recent cannabis use by blood toxicology or self-report. Used Newcastle-Ottawa scale for bias assessment and random effects models.
What this study cannot tell us
High heterogeneity between studies. Observational designs cannot prove causation. Different studies used different measures of cannabis exposure (blood THC vs self-report). Publication bias was possible. The analysis could not determine a dose-response relationship.
How to read the evidence
Systematic review and meta-analysis with comprehensive search strategy. Strong methodology despite limitations inherent to the observational studies included.
When this study was published
Published in 2012. Subsequent meta-analyses have generally confirmed an increased crash risk, though estimates vary.
The bigger picture
This meta-analysis provided the quantitative foundation for drug-impaired driving policies. The finding that cannabis approximately doubles crash risk, while meaningful, is notably smaller than alcohol impairment, which typically increases risk 5-20 fold depending on blood alcohol level.
Questions still open
- What blood THC level corresponds to meaningful impairment? How does cannabis-impaired driving risk compare to other common impairments like fatigue? Does tolerance in regular users reduce driving risk?
Common questions
How much does cannabis increase crash risk?
Is driving after cannabis as dangerous as drunk driving?
Read the original research
Acute cannabis consumption and motor vehicle collision risk: systematic review of observational studies and meta-analysis.
BMJ (Clinical research ed.), 344, e536
Citation
Asbridge, Mark; Hayden, Jill A; Cartwright, Jennifer L. (2012). Acute cannabis consumption and motor vehicle collision risk: systematic review of observational studies and meta-analysis.. BMJ (Clinical research ed.), 344, e536. https://doi.org/10.1136/bmj.e536