To investigate the predictors of persistent driving risk related to sleepiness in patients with obstructive sleep apnea syndrome treated by continuous positive airway pressure. Longitudinal analysis of a prospective national database including 5,308 patients with obstructive sleep apnea syndrome and an indication of continuous positive airway pressure. Near-misses related to sleepiness, accidents related to sleepiness, and sleepiness at the wheel were assessed before initiation and after ≥ 90 days of treatment. Multivariable associations with the cumulative incidence of near-misses and accidents under treatment were calculated using Cox models adjusted for age, sex, obesity, sleep duration, sleepiness at the wheel, accidents/near-misses history, depressive symptoms, residual apnea-hypopnea index, and adherence to treatment. Residual sleepiness at the wheel under treatment was associated with eight-fold higher incidence of near-misses related to sleepiness (HR=8.63 [6.08-12.2]) and five-fold higher incidence of accidents related to sleepiness (HR=5.24 [2.81-9.78]). Adherence ≤4h/night was also a significant predictor of persistent driving risk (HR=1.74 [1.12-2.71] for near-misses and HR=3.20 [1.37-7.49] for accidents). Residual sleepiness at the wheel and treatment-adherence ≤4h/night are easy-to assess markers to detect persistent driving risk during the follow-up evaluations of patients under treatment. Health professionals, but also policy makers, should be aware of the crucial importance to systematically evaluate these elements during the follow-up evaluations of the patients with obstructive sleep apnea syndrome treated by continuous positive airway pressure to better evaluate their driving risk.