Abstract: For the problem of great sampling number and low efficiency of Monte Carlo method, the sampling issue is solved by using control variate method, which has fast convergence feature when evaluating reliability of power system with large scale wind farm by considering such characteristics as small unit capacity and large unit number of large wind farm. In this paper, the reliability index of conventional units is taken as control variable, the analytical method is used to calculate their reliability index, and sampling of conventional units and wind turbine units is carried out successively and circularly. By taking the IEEE RTS79 test system as example, the proposed algorithm is verified and is compared with the common Monte Carlo method and the average and scattered sampling method, and the factors that affect sampling efficiency are analyzed. The simulation results show that the proposed method can largely reduce the number of sampling and improve sampling efficiency under certain precision.