Analyses of accumulated fatigue level for long-distance coach drivers are considered as one effective way for traffic safety.Field experiments are conduct among drivers from Helen City,Heilongjiang.The driving video and measurements on perception j udgment ability,self-evaluation of fatigue level and other indicators are collected.The PER-CLOS-P80 is extracted from videos using analysis and processing software,which then used as the benchmark of fatigue level.Rules of fatigue accumulation for long-distance coach drivers are analyzed from two aspects:original fatigue and driving fatigue.The results show that original fatigue is of negatively correlation with sleeping hours in the day before driving and is of positive correlation with driving duration.Meanwhile,the driving fatigue is of positively correlation with original fatigue and continuous driving duration and is of negative correlation with cumulative resting time.A regression model of fatigue accumulation based on driving and resting time of drivers is established.Regression methods are used to calibrate the model and the calibrated model is with goodness of fit of 0.929.The fatigue accumulation model proposed in this study demonstrated the threshold of continuous driving time (when F=0.3)with different original fatigue values. Suggestions and recommendation could also be provided to commercial transportation companies or drivers based on the regression model for optimizing their operation schedule to reduce the risk of fatigue driving.