Background: Psychomotor studies suggest that commonly prescribed psychoactive drugs impair driving skills. We have examined the association between the use of psychoactive drugs and road-traffic accidents. Methods: We used dispensed prescribing as a measure of exposure in a within-person case-crossover study of drivers aged 18 years and over, resident in Tayside, UK, who experienced a first road-traffic accident between Aug 1, 1992, and June 30, 1995, and had used a psychoactive drug (tricyclic antidepressant, benzodiazepine, selective serotonin-reuptake inhibitor, or other psychoactive drug [mainly major tranquillisers]) between Aug 1, 1992, and the date of the accident. For each driver, the risks of having a road-traffic accident while exposed and not exposed to a drug were compared. Findings: 19,386 drivers were involved in a first road-traffic accident during the study period. 1731 were users of any study drug. On the day of the accident, 189 individuals were taking tricyclic antidepressants (within-patient exposure odds ratio for an accident 0.93 [95% CI 0.72-1.21]), 84 selective serotonin-reuptake inhibitors (0.85 [0.55-1.33]), 235 benzodiazepines (1.62 [1.24-2.12]), and 47 other psychoactive drugs (0.88 [0.62-1.25]). The risk associated with benzodiazepine use decreased with increasing driver's age and was greater when the breath test for alcohol was positive. A dose-response relation was evident with benzodiazepines. The increased risk with benzodiazepines was significant for long-half-life drugs, used as anxiolytics, and for short-half-life hypnotics (all zopiclone). Interpretation: Users of anxiolytic benzodiazepines and zopiclone were at increased risk of experiencing a road-traffic accident. Users of anxiolytic benzodiazepines and zopiclone should be advised not to drive.