In order to solve the problems of single evaluation criteria and poor environmental adaptability of automobile anti-collision early warning system in complex environment, in this paper, under the influence of many factors, the collision risk identification methods were studied, to establish "driver – vehicle - road - environment" synergy of vehicle collision risk situation multi-level comprehensive evaluation, the vehicle collision risk situation assessment under the influence of many factors as a mixture of multiple criteria decision making (MCDM) problems, put forward a kind of combination weight TOPSIS method based on game theory method of vehicle collision risk situation assessment. Firstly, the collision risk situation assessment system architecture was established based on the vehicle-road cooperation technology, and the subjective and objective weights of the evaluation indexes were determined by the Delphi method and entropy method, and the optimal equilibrium solution between the weights of the selected criteria was obtained by the game theory method. Secondly, TOPSIS method was applied to determine the distance between the current driving state and the driving state in the decision-making information table, and to match the evaluation level of collision risk. Finally, Prescan software was used to build simulation scenarios for simulation analysis. The result shows that this method can integrate multiple factors related to driving safety in complex environments to evaluate collision risk, providing more accurate decision-making for automobile collision prevention system.