The main symptom of drivers with obstructive sleep apnea hyperpnoea syndrome (OSAHS)commonly have excessive daytime sleepiness,which would increase the probability of mis-operations that may result in accidents.As a result,it is crucial to identify the physiological status of OSAHS drivers in order to avoid or reduce the probability of accidents.The brain waves,which can reflect the physiological changes of OSAHS drivers,are used to identify the driv-ing state in this paper.The indicators for early warning of accidents are identified.20 drivers with OSAHS are recruited to drive in a driving simulator while detecting their brain waves.The driving environment and requirements,such as prohibi-tions of phone calls and over speed,etc.,are identical for all of the drivers.The values ofα andβ rhythm of EEG as well as accident types and occurrence times are recorded during the simulations.The fluctuations of brain waves under different driving conditions are analyzed from both macro (waveforms)and micro (data)perspectives.The results indicate that theα rhythm is either suppressed or vanished when an accident occurs.The mean value ofβ/(α+β)during an accident has significant distinctions with which in regular driving modes.As a result,both the change ofα rhythm and the average val-ue ofβ/(α+β)can be recognized as early warning indicators of sleepiness for OSAHS drivers during driving.Further-more,the average value ofβ/(α+β)in 30 s before the accidents can be treated as a threshold value of early warning of accidents.