Abstract: Road accidents happened frequently due to the risky riding of take-away riders. In order to explore the complex mechanism and evolutionary trends of take-away riders’ decision-making regarding risky riding behavior, taking Meituan take-away riders as the research object, an evolutionary game model of the platform and riders was constructed. The stability conditions of strategy selection for both game parties were systematically analyzed. Based on parameter assumptions of the game model, numerical simulation analysis was conducted to investigate the dynamic mechanisms of platform performance reward and violation penalty on take-away riders’ risky riding behaviors. The results show that the platform performance reward has a significant impact on take-away riders’ choice of risky riding behavior, higher performance rewards increase the probability of take-away riders’ choice of risky riding. Platform violation penalty has a certain inhibitory effect on restraining take-away riders' risky riding behavior. Setting reasonable violation penalty can reduce the probability of risky riding. However, a substantial increase in violation penalty amount may cause fluctuations in the evolution of risky riding behavior, and reduce the probability of strict supervisory by platform. Moreover, reducing regulatory costs can effectively increase the platform's willingness to enforce strict supervision, and improve the probability of take-away riders’ choice of safe riding. Therefore, the platforms can design multi-dimensional performance reward mechanisms based on delivery efficiency and safety assurance to guide and encourage riders strictly complying with traffic regulations. By adjusting the severity of violation penalty, it can avoid irrational behaviors where take-away riders choose retaliatory dangerous riding due to excessive penalty. By increasing the platform’s willingness to strictly supervise and the perceived cost of take-away riders’ dangerous riding, the platform can constrain take-away riders to adopt more safe riding strategies.