We ran a randomized cross-over design study under sleep-deprived and non-sleep-deprived driving conditions to test the effects of sleep restriction on real driving performance. The study was performed in a sleep laboratory and on an open French highway. Twenty-two healthy male subjects (age = 21.5 +/- 2 years; distance driven per year = 12,225 +/- 4739 km (7641 +/- 2962 miles) [mean +/- S.D.]) drove 1000 km (625 miles) over 10 h during five 105 min sessions on an open highway. Self-rated fatigue and sleepiness before each session, number of inappropriate line crossings from video recordings and simple reaction time (RT) were measured. Total crossings increased after sleep restriction (535 crossings in the sleep-restricted condition versus 66 after non-restricted sleep (incidence rate ratio (IRR): 8.1; 95% confidence interval (95% CI): 3.2-20.5; p < 0.001)), from the first driving session. The interaction between the two factors (conditionxtime of day) was also significant (F(5, 105) = 3.229; p < 0.05). Increasing sleepiness score was associated with increasing crossings during the next driving session in the sleep-restricted (IRR: 1.9; 95% CI: 1.4-2.4) but not in the non-restricted condition (IRR: 1.0; 95% CI: 0.8-1.3). Increasing self-perceived fatigue was not associated with increasing crossings in either condition (IRR: 0.95; 95% CI: 0.93-0.98 and IRR: 1.0; 95% CI: 0.98-1.02). Rested subjects drove 1000 km with four shorts breaks with only a minor performance decrease. Sleep restriction induced important performance degradation even though time awake (8h) and session driving times (105 min) were relatively short. Major inter-individual differences were observed under sleep restriction. Performance degradation was associated with sleepiness and not fatigue. Sleepiness combined with fatigue significantly affected RT. Road safety campaigns should encourage drivers to avoid driving after sleep restriction, even on relatively short trips especially if they feel sleepy.