Marijuana-impaired driving is a growing public safety concern. After alcohol, cannabis is the substance most commonly detected in impaired drivers, and its prevalence has increased substantially following legalization in multiple states. Yet the regulatory and enforcement response to cannabis-impaired driving has not kept pace with the growth of legal cannabis, in part because cannabis impairment presents unique challenges that make the policy tools developed for alcohol-impaired driving difficult to apply directly. This article examines what marijuana actually does to driving, how the law addresses it, why detection is so difficult, and what the crash risk evidence shows.
How Marijuana Impairs Driving
Safe driving requires sustained attention, rapid response to changing conditions, precise motor coordination, lane and speed management, divided attention across multiple information streams, and good judgment under uncertainty. Cannabis impairs all of these through THC’s effects on the brain’s prefrontal cortex, cerebellum, basal ganglia, and hippocampus.
Controlled studies and driving simulator research have documented the following impairments from cannabis intoxication:
Reaction time. THC slows reaction time to unexpected events (braking response time, response to traffic signals, and response to pedestrians) in a dose-dependent manner. This is among the most consistent findings across laboratory, simulator, and on-road studies.
Lane tracking (lateral position control). Cannabis impairs the ability to maintain consistent lateral position within a lane, a measure called standard deviation of lateral position (SDLP). This impairment is dose-dependent and is the most reliably detected cannabis driving impairment in simulator studies.
Divided attention. Driving requires simultaneous monitoring of multiple information sources: road conditions, mirrors, signals, speed, other vehicles. Cannabis specifically impairs divided attention, the ability to process multiple information streams simultaneously. A 2015 study found THC hurts the driver’s ability to multitask.
Speed and headway regulation. Cannabis-using drivers in simulator studies tend to drive slower and maintain greater headway. The NHTSA 2017 Report to Congress noted this pattern: subjects dosed on marijuana showed reduced mean speeds, increased following distance, and more time below the speed limit. This contrasts with alcohol, which tends to produce faster driving and shorter following distances. This compensation behavior is one reason cannabis-impaired crash risk is harder to detect than alcohol-impaired risk in naturalistic data.
Working memory and executive function. Decision-making, route planning, navigation, and judgment under uncertainty are all impaired. THC-related working memory deficits can cause drivers to miss turns, misread conditions, or fail to process sudden changes in traffic patterns.
Edible cannabis: delayed onset, longer duration. A particularly dangerous driving context is edible cannabis use. Edibles have delayed onset (up to 2 hours), meaning drivers may underestimate their impairment at the time of driving; and longer duration (6–12+ hours), meaning impairment may persist well into periods when the driver believes it has worn off.
Cannabis vs. Alcohol Impairment: Major Differences
Cannabis and alcohol both impair driving, but through different mechanisms and with different behavioral signatures:
The Crash Risk Evidence
Multiple meta-analyses have found that recent cannabis use increases crash risk. A systematic review by Brubacher et al. identified meta-analytic estimates of crash risk ranging from ORs of 1.36 to 2.66 for recent cannabis use. A 2022 NEJM study by Brubacher and colleagues in British Columbia found significantly increased crash responsibility in cannabis-positive drivers, with the risk increasing at higher THC blood concentrations.
However, the crash risk evidence is more complicated than these summary figures suggest:
The demographic confound. Cannabis users are disproportionately young men, who have the highest baseline crash risk of any demographic group. NHTSA’s large case-control study found that once age, sex, and alcohol use were controlled for, drivers testing positive for THC showed no statistically significant increase in crash risk compared to drug-free drivers. This finding is contested and does not mean cannabis is safe to drive on; rather, it illustrates how difficult it is to isolate the cannabis-specific contribution to crash risk in naturalistic driving data.
THC detection window vs. impairment window. THC remains detectable in blood for hours to days after impairment has cleared, and in urine for days to weeks. Studies using any detectable THC as the exposure variable are capturing both impaired and non-impaired drivers. Studies that use higher blood THC thresholds (e.g., ≥5 ng/mL whole blood) find stronger crash risk associations than those using any detectable THC.
Colorado real-world data. Post-legalization state-level data illustrates the practical scale of the problem. In Colorado, 32% of roadway fatalities in 2024 involved a driver suspected to be cannabis-impaired at the time of the crash. This is not a small or marginal public safety problem.
State DUI Laws for Cannabis
All 50 states and DC make it illegal to drive under the influence of cannabis. The approach differs significantly by state:
Impairment-based (effects-based) laws. The majority of states use impairment-based standards: the prosecution must demonstrate that the driver was actually impaired at the time of driving, based on evidence including officer observations (behavior, FST performance, Drug Recognition Expert assessment), driving pattern, and any available chemical test results. No specific THC concentration must be proved, but impairment must be established.
Per se THC limits. Several states have established per se THC blood concentration limits, most commonly 5 nanograms per milliliter (ng/mL) of whole blood (including Colorado, Montana, Nevada, Ohio, and Pennsylvania). A driver testing at or above this level is per se impaired, similar to the 0.08% BAC standard for alcohol, without requiring additional proof of behavioral impairment.
Zero-tolerance laws. Eleven states (including Arizona, Delaware, Georgia, Indiana, Iowa, Michigan, Oklahoma, Rhode Island, South Dakota, Utah, and Wisconsin) maintain zero-tolerance laws for one or more cannabis compounds in blood, making any detectable THC or metabolite illegal while driving. These laws have been criticized for ensnaring drivers who are not actively impaired.
The critical policy problem with per se and zero-tolerance approaches is that, unlike alcohol, blood THC concentration does not reliably predict impairment in an individual. This is because:
- Chronic heavy cannabis users develop tolerance and may show high blood THC levels without the impairment that the same level would produce in an infrequent user
- THC concentrations fall rapidly after smoking, but impairment may outlast detectable THC, while residual THC may be detectable in blood after impairment has resolved
- Regular cannabis users (including medical patients) may maintain blood THC levels above per se limits even after extended abstinence, not because they are impaired but because THC is released from fat stores
The AAA Foundation for Traffic Safety evaluated THC-positive drivers and drug recognition expert assessments and concluded that the data did not support any fixed per se THC limit as a reliable predictor of impairment. As AAA Director of Traffic Safety Advocacy and Research Jake Nelson explained, there is no concentration of THC that allows us to reliably predict impairment behind the wheel in the way we can with alcohol.
The Detection Problem
The absence of a reliable impairment-linked THC threshold is compounded by the state of roadside detection technology:
No equivalent of the breathalyzer. Alcohol can be accurately measured from breath, quickly, roadside, with a widely validated and legally defensible device. No equivalent device exists for THC. Blood testing is the most accurate method for measuring THC, but it requires a trained professional, a blood draw, laboratory analysis, and often days to weeks to return results. This delay makes blood testing impractical as a primary enforcement tool.
Oral fluid testing: emerging but imperfect. Oral fluid (saliva) devices have been developed and are in use in some jurisdictions. They can detect the presence of THC in oral fluid relatively quickly. However, THC concentration in oral fluid correlates poorly with blood THC level and even more poorly with actual impairment. They are useful for detection but not for quantifying impairment. Cross-contamination (from cannabis smoke in the environment) and variability in absorption reduce reliability.
Drug Recognition Experts (DREs). Law enforcement trains Drug Recognition Experts to identify signs of impairment from specific drug categories through a standardized 12-step evaluation protocol. DRE evaluation is time-consuming and resource-intensive, but it is currently one of the better tools available for detecting cannabis-specific impairment rather than merely cannabis presence. DRE evaluations are admissible in many states but require trained officers who are not universally available.
Urine testing. Urine THC testing detects the cannabis metabolite THC-COOH, not active THC. THC-COOH can remain detectable for days to weeks after use, providing no meaningful information about whether the driver was impaired at the time of driving. Urine testing is largely useless for determining impairment and is generally not used in roadside DUI enforcement for this reason.
What Drivers Need to Know
Several practical points follow from this evidence:
- Marijuana is illegal to drive under the influence of in every state, regardless of legal status
- The claim that cannabis makes some people better drivers is not supported by controlled evidence
- Edibles are particularly dangerous for driving because of delayed onset: users may feel unimpaired and begin driving before peak impairment arrives
- There is no reliable way to self-assess cannabis impairment for driving purposes. Cannabis users are particularly poor at accurately assessing their own driving impairment
- In states with per se limits, a blood THC level at or above the threshold constitutes legal impairment regardless of how the driver feels
- Driving while under the influence of cannabis can result in DUI arrest, criminal charges, license suspension, increased insurance rates, and civil liability for crashes
Professional Perspective
The prospective study of cannabis use as a crash risk factor requires careful methodological attention to distinguishing impairment from detection. Much of the literature has been hampered by use of metabolites rather than active THC as exposure measures, and by the failure to control for the demographic confounders that are especially pronounced in cannabis-using driving populations.
“Controlled experiments show that cannabis impairs the psychomotor skills required for safe driving.”
For people with cannabis use disorder who drive regularly, the driving safety implications of their use are a concrete and immediate harm that treatment directly addresses. Cannabis cessation eliminates cannabis-impaired driving risk. The difficulty many people have accurately assessing their own cannabis impairment means that even users who believe they drive safely when using cannabis may be objectively impaired. Treatment programs that address CUD should include education on cannabis-impaired driving as part of the harm reduction conversation, particularly in legal states where the perception that cannabis use is safe has broadly reduced risk perception among drivers.