Experimental studies have repeatedly shown that thc impairs cognition, psychomotor function and actual driving performance in a dose-related manner. The degree of performance impairment observed in experimental studies after doses up to 300 μ g/kg thc were equivalent to the impairing effect of an alcohol dose producing a blood alcohol concentration ⪰ 0.05 g/dl, the legal limit in most european countries. Higher doses of thc have been shown to produce even larger impairment. Highly automated behaviors were more affected by thc compared to more complex driving tasks requiring conscious control. Epidemiological findings on the role of thc in vehicle crashes have sometimes contrasted findings from experimental research. Case-control studies generally confirmed experimental data, but culpability surveys showed little evidence that crashed drivers who only used cannabis are more likely to cause accidents than drug-free drivers. However, most culpability surveys have established cannabis use by determining the presence of an inactive metabolite of thc that can be detected for days after smoking and can only be taken as evidence for past use of cannabis. Surveys that established recent use of cannabis by directly measuring thc in blood showed that thc positives, particularly at higher doses, are about 2–7 times more likely to be responsible for their crash compared to drivers who had not used drugs or alcohol. Together these epidemiological data indicate that recent use of cannabis increases crash risk, whereas past use of cannabis does not. Experimental and culpability studies have demonstrated that performance impairment and crash risk increase as a function of thc concentration. Moreover, performance impairments as assessed during acute thc intoxication in experimental performance studies were highly correlated to thc-induced crash risk in culpability studies. Experimental and epidemiological research provided similar findings concerning the combined use of thc and alcohol in traffic.keywordscrash riskaccid analkdyh ehhqactual driving performancedvvrfldwhg zlwkthese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.