Though fatigue and sleepiness at the wheel are well-known risk factors for traffic accidents, many drivers combine extended driving and sleep deprivation. Fatigue-related accidents occur mainly at night but there is no experimental data available to determine if the duration of prior driving affects driving performance at night. Participants drove in 3 nocturnal driving sessions (3-5 am, 1-5 am and 9 pm-5 am) on open highway. Fourteen young healthy men (mean age [+/-SD] = 23.4 [+/-1.7] years) participated Inappropriate line crossings (ILC) in the last hour of driving of each session, sleep variables, self-perceived fatigue and sleepiness were measured. Compared to the short (3-5 am) driving session, the incidence rate ratio of inappropriate line crossings increased by 2.6 (95% CI, 1.1 to 6.0; P<.05) for the intermediate (1-5 am) driving session and by 4.0 (CI, 1.7 to 9.4; P<.001) for the long (9 pm-5 am) driving session. Compared to the reference session (9-10 pm), the incidence rate ratio of inappropriate line crossings were 6.0 (95% CI, 2.3 to 15.5; P<.001), 15.4 (CI, 4.6 to 51.5; P<.001) and 24.3 (CI, 7.4 to 79.5; P<.001), respectively, for the three different durations of driving. Self-rated fatigue and sleepiness scores were both positively correlated to driving impairment in the intermediate and long duration sessions (P<.05) and increased significantly during the nocturnal driving sessions compared to the reference session (P<.01). At night, extended driving impairs driving performances and therefore should be limited.