In order to analyze the impact of variations of key parameters， such as load model and fault time etc.， on power system dynamic stability， this paper applies a new method—the stochastic response surface method（SRSM） to analyze uncertainties of these parameters. Polynomial expressions between parameters and responses of system can be established by this method. Compared with the traditional approaches， the presented method can obtain statistical information with less simulation time. The case study on the IEEE 9-bus system demonstrates the validity of SRSM. Simulation results on the IEEE 39-bus system show that the uncertainties from key parameters， especially the different load model structures， have great impact on power system dynamic stability， which is directly related to the security and stability of power systems.