Patients with obstructive sleep apnoea (OSA) are reported to have an increased risk of road traffic accidents. This study examines the nature of the impairment during simulated steering in patients with OSA, compared to normal subjects following either sleep deprivation or alcohol ingestion. Twenty-six patients with OSA and 12 normal subjects, either deprived of one night's sleep or following alcohol ingestion [mean (SD) alcohol blood level 71.6 mg dl(-1) (19.6)], performed a simulated steering task for a total of 90 min. Performance was measured using the tendency to wander (SD), deterioration across the task, number of 'off-road' events and the reaction time to peripheral events. Control data for OSA, sleep deprivation and alcohol were obtained following treatment with nasal continuous positive airway pressure (nCPAP), after a normal night of sleep, and following no alcohol, respectively. Patients with untreated OSA, and sleep-deprived or alcohol-intoxicated normal subjects performed significantly less well, compared to their respective controls (P<0.01 for all tests), with untreated OSA lying between that of alcohol intoxication and sleep deprivation. Alcohol impaired steering error equally throughout the whole drive, whilst sleep deprivation caused progressive deterioration through the drive, but not initially. Untreated OSA was more like sleep deprivation than alcohol, although there was a wide spread of data. This suggests that the driving impairment in patients with OSA is more compatible with sleep deprivation or fragmentation as the cause, rather than abnormal cognitive or motor skills.