THC and driving: The simulator crashes peaked at 4 hours, not right after

In this double-blind crossover RCT (randomized controlled trial where each person receives each condition), inhaled THC increased driving simulator collisions, with the clearest increase observed 4 hours after 10 mg and 30 mg THC. Higher THC Cmax (peak blood THC level) was strongly linked to more collisions at 4 hours. Higher self-rated sleepiness on the Karolinska Sleepiness Scale (a 1-9 rating of how sleepy someone feels) was also linked to higher collision rates on the simulator.

This was a randomized, double-blind, crossover RCT, meaning the same 30 participants completed three separate sessions under placebo, 10 mg THC, and 30 mg THC, and neither participants nor staff knew the condition at the time. Each session was separated by a 7-day washout period, and participants stayed in the hospital for 24 hours while blood samples were collected. Driving performance was tested on the York driving simulator, and participants also completed VAS (visual analog scale) ratings and the Karolinska Sleepiness Scale to capture how they felt during testing. The most important weakness is that simulator collisions are not the same as real-world crashes, and the sample was small and limited to males aged 18-30.

By 2025, policy and safety debates increasingly focused on whether blood THC can serve as a meaningful marker for driving risk, similar to blood alcohol. This study targeted a practical question for traffic safety science: whether a controlled THC exposure changes crash-like outcomes, and whether subjective ratings (VAS, or visual analog scales) track real performance in the same session. The crossover setup (the same participants under placebo, 10 mg, and 30 mg) reduces between-person differences that often complicate real-world crash research.

A headline might read as if “THC increases crashes,” but the timing detail matters here because the strongest increase was reported at 4 hours after inhalation in this simulator protocol, not immediately after dosing. The study also sits at the messy intersection of two ideas that often get treated as the same thing: THC dose (10 mg vs 30 mg) is something the trial controlled, while THC Cmax (peak blood THC) is a biological measure that varied across people and was linked to collisions (IRR = 37.7) in this small sample. The abstract’s claim of no collision difference between occasional and chronic users runs against the popular assumption that frequent users are protected by tolerance, but the abstract does not report the subgroup effect sizes, so the strength of that null finding cannot be checked here. Because every participant had at least a year of cannabis use, these simulator results do not speak to first-time users or cannabis-naive drivers.

Only 30 people were studied, all male and ages 18-30, which leaves open whether the same pattern holds in other groups. The outcome was crashes in a driving simulator, which can model driving errors but does not capture real-road complexity, risk-taking, or consequences. Chronic versus occasional use was discussed, but the abstract does not provide the underlying subgroup statistics, making the tolerance comparison hard to judge from the abstract alone.