Abstract: According to the variability characteristics of fatigue,based on indexes of heart rate variability( HRV),a method of recognizing driver's mental fatigue was proposed. The method proposed hierarchy partition of driving mental fatigue through using the driver's behavior performance as objective evaluation indexes,which could be the response on the variability of fatigue state. Meanwhile,according to the 3time-domain indexes and 5 frequency-domain indexes of HRV as recognizing fatigue characterized factors and which combined with support vector machine( SVM),the model was established to recogniz the state of driver's mental fatigue. Finally,combining with the examples,the mental fatigue was divided into two classifications. 4 hours of continual driving behavior performance and ECG data from 10 drivers were collected to test the model. The result showed that the average recognition accuracy rate was between70%- 82%,and the average accuracy rate was 75%.