OBJECTIVE: To examine the relationship between benzodiazepine (BZD) use and motor vehicle accidents (MVAs). DATA SOURCES: MEDLINE was searched from 1980 to 1997 using the key words traffic accidents or motor vehicle accidents and benzodiazepines (and alternative terms and outcomes) in English, German, French, or Italian. STUDY SELECTION: Case-control studies of BZDs and MVAs; police or emergency studies of BZD use among travelers; driving tests with subjects taking BZDs. Outcomes were impaired driving, accidents; mortality; postaccident medical attention, emergency ward care, or hospitalization. Quality criteria were whether all driving BZD users and non-users had an equal chance of entering the study; whether medication dosage and timing were ascertained; whether all kilometres driven by BZD users and non-users were studied; whether all types of accidents were ascertained; and whether medical conditions were controlled for. SYNTHESIS: In case-control studies, the odds ratios for mortality and emergency medical treatment ranged from 1.45 to 2.4 in relation to time of use and quantity of drug taken. In police and emergency ward studies, BZD use was a factor in 1% to 65% of accidents (usually 5% to 10%). In two studies where subjects had blood alcohol concentrations less than the legal limit, BZDs were found in 43% and 65% of subjects. In one study with controls, 5% of drivers and 2% of controls in accidents had used BZDs. CONCLUSIONS: Case-control studies suggest using BZDs approximately doubles the risk of motor vehicle accidents. The risk for drivers older than 65 of being involved in reported motor vehicle collisions is higher when they take longer-acting and larger quantities of BZDs.