Abstract: While existing studies are mostly concerned about driver fatigue recognition algorithm, this paper proposes a quantitative model for performance analysis of driver fatigue monitoring devices, which is suitable for analyzing the role of the device in traffic accidents prevention. First, we constructed a model of the accident process of the vehicle equipped with a driver fatigue monitoring device based on continuous Markov Process. The Mean Time to Failure (MTTF) was taken as the measurement of the devices' performance. Then we obtained the relationship between MTTF and the performance indicators of driver fatigue monitoring devices by solving the model. Finally, several suggestions were presented for the application and development of driver fatigue monitoring devices. This model can be applied to analyze real problems directly after parameter calibration.