Abstract: To study the effects of secondary task on driving safety, data by driving simulator regarding vehicle operation states under normal and secondary task driving conditions were collected and analyzed. Based on the principle of significant difference, parameters such as speed, acceleration and steering wheel angle were applied to formulate the evaluation indicator system. With the optimized target to reflect the most information utilizing least factors, 3 kinds of factor variables were abstracted by means of factor analysis to ascertain the impact load coefficients on factors of different evaluation parameters. Therefore, a driving safety evaluation model based on vehicle operation states was formulated. At last, combining experiment data and driver type, the relationship between evaluated score and safety level was established. Experiment results show that the model built in this study could estimate driving safety level in different driving conditions and the model provides an effective method for driving safety monitoring and early warning.