In order to study the dynamic performance and parameter optimization of fully flexible railway freight cars, taking C80 railway freight cars as the research object, the dynamic model of C80 railway freight cars was established by joint simulation of NASTRAN and SIMPACK. The model is verified by experimental modal and dynamic calculation results, in which the maximum error of mode is 9.85%, and the maximum error of derailment coefficient and wheel load reduction rate is 6.0% and 9.5% respectively, which verifies the correctness of the model. Based on the dynamic model, the dynamic performance of empty and heavy cars in the curve section at different speeds was calculated; the DOE(design of experiments) experimental group was established, and the proxy model of derailment coefficient, lateral stationarity and vertical stationarity index of C80 railway freight car was established by response surface method, and the running parameters such as running speed, track curve radius and load were optimized based on multi-island genetic algorithm. The results show that the overturning coefficient and lateral smoothness of empty and heavy vehicles increase with the increase of speed, and the overturning coefficient of empty vehicles is lower than that of heavy vehicles. According to the design requirements of the dynamic performance of C80 railway freight cars, the dynamic performance can be effectively improved by adjusting the running speed, load and track curve radius. The maximum relative errors between the optimization results and the real values of derailment coefficient, lateral stationarity and vertical stationarity are 6.74%, 0.20% and 2.10%, respectively, which meet the accuracy requirements of the model. The study of flexible modeling and dynamic performance parameter optimization of C80 railway freight car provides an important reference for its engineering application in heavy-haul railway freight car field.