ObjectiveThe aim of the present research was to examine associations between poor driving behaviour (DB), driving when fatigued (DF), risk taking (RT) and road traffic accidents (RTAs).DesignThe study involved a cross-sectional online survey of clients of an insurance company. The survey measured DB (speeding, distraction, lapses of attention and aggression), RT and frequency of driving when fatigued (DF, driving late at night, prolonged driving, driving after a demanding working day and driving with a cold). Demographic, lifestyle, job characteristics and psychosocial factors were also measured and used as covariates.SettingCardiff, UK.Sample3000 clients of an insurance company agreed to participate in the study, and 2856 completed the survey (68% woman, 32% man; mean age: 34 years, range 18–74 years).Main outcome measuresThe outcomes were RTAs (requiring medical attention; not requiring medical attention), where the person was the driver.ResultsFactor analyses showed that DB, RT and fatigue loaded on independent factors. Logistic regressions showed that poor DB, frequently DF and taking risks predicted medical and non-medical RTAs. These effects were additive and those who reported poor DB, driving when fatigue and taking risks were twice as likely to have an RTA. These effects remained significant when demographic, lifestyle, medical, driving, work and psychosocial factors were covaried.ConclusionsPoor DB, DF and RT predict RTAs. There are now short measuring instruments that can assess these, and driver education programmes must increase awareness of these risk factors.