In order to reduce the incidence of traffic accidents and improve the traffic safety of mountainous expressway, the influencing factors of the number of vehicles involved in traffic accidents are analyzed. Firstly, this paper analyzed the time-space characteristics and causes of more than 10000 traffic accidents in the past eight years of a certain expressway in Western China; secondly, taking the number of vehicles involved in the accident as the dependent variable, it was divided into three levels: single vehicle, double vehicle and multi vehicle accidents. Eight potential independent variables were selected as the analysis factors of ordered logit model from the aspects of driver, vehicle and driving environment, and six independent variables with significance less than 0.05 were obtained; Finally, six significant variables were classified by using ordered logit model, and their classification odds ratio (or) was obtained through analysis, and the correctness of the independent variable classification risk was verified by using negative binomial regression (NB) model. The analysis results show that the advantage ratio of winter, involving large cargo, insufficient spacing, toll station, sunny day and other categories is the largest, and the advantage ratio changes more in the analysis of the ordered logit model, which shows that the ordered logit model is more suitable than the negative binomial regression model to analyze the classification advantage ratio of the number of vehicles involved in accidents. The conclusion shows that controlling the factors with large advantage ratio can reduce the number of accident vehicles, thus indirectly reduce the incidence of serious accidents, which provides a strong decision basis for highway management.