Wind speeds of different wind turbines in a big wind farm always have correlations because of their close geographical location. This paper applies stochastic response surface method (SRSM) to analyze transient stability of power system. This method establishes polynomial relationship between inputs and transient stochastic output responses, and estimates the probability statistics of output responses by less times of simulation. Then this paper builds wind speed model by t-Copula function and uses rank correlation matrix to measure the correlations of wind speeds. Wind speed vector is transformed as the input of SRSM in order to study the probability statistics of output responses. The case study on Yunnan Power Grid shows that correlations of wind speeds have certain effect on the transient stability of power system. Taking the correlations of wind speeds into consideration makes the analysis of system’s transient stability more practical, and can guide the stable operation of systems.