The randomicity of wind power caused serious influence on safe operation of power system, and the spare capacity is deemed as an effective way to reduce risk produced by wind power. On the premise of complete acceptance of wind power, the probability characteristic of wind speed and load are taken into consideration to count spinning reserve capacity under a certain confidence level and the reserve capacity for accident is counted by 10 percent of maximum load. Moreover, Bootstrap test is used to transform the chance constraints into deterministic ones to improve precision. Generating and reserve cost are added to objective function, the stochastic optimization model for combined wind and thermal power system is built. And furthermore, integer variable is simplified to reduce the degree of model solving difficulty. Finally, the allocation of generation and reserve tasks among thermal power plants and economy under different confidence levels are analyzed.