To analyze various factors influencing the accident severity of urban river-crossing tunnels, 12 influence factors were selected according to the three traffic elements of vehicle, road, and environment, including accident-involved vehicle type, tunnel length, tunnel speed limit, accident occurrence time, and weather. These factors were based on the historical data of 14 urban river-crossing tunnels in Shanghai. A binary logistic regression model was also utilized to identify significant influence factors and analyze the influence degrees of their accident severity. Significance analysis reveals that accident occurrence place, accident-involved vehicle type, tunnel speed limit, number of vehicles involved, and accident type are significantly correlated with accident severity. Moreover, VISSIM traffic simulation model and conflict simulation analysis software called surrogate safety assessment model were utilized to simulate different urban rivercrossing tunnel scenes. The simulation explored the influence degrees of traffic volume, tunnel longitudinal slope, tunnel speed limit, and ratio of heavy trucks, which stand for three traffic elements (i.e., vehicle, road, and environment, respectively) on accident severity. Results show that the factors exerting the greatest influences on accident severity are traffic volume, number of vehicles involved, accident occurrence place, and accident type, followed by tunnel longitudinal slope, tunnel speed limit, and accident-involved vehicle type. Traffic volume, number of vehicles involved, accident type, tunnel longitudinal slope, and accident-involved vehicle type have a significantly positive correlation with accident severity, whereas accident occurrence position and tunnel speed limit have a significantly negative correlation with accident severity. The study results provide a new idea for researches on urban river-crossing tunnels and provide basis and method for tunnel entrance design, speed limit adjustment and regulation of traffic volume.