A new model and algorithm for reactive power optimization of distribution networks embedded with multiple wind power generators（WPGs） are presented to overcome the defects of conventional methods. The scenario analysis method is applied to describe the random outputs of single and multiple WPGs of different parameters. A scenario partition rule is established for the distribution networks embedded with multiple WPGs of different parameters， and a scenario model is built for reactive power optimization of these distribution networks. An efficient algorithm is presented to solve the newly built scenario model. In this algorithm， the order of disposal of system scenarios are determined by their similarity， and the load flow results of a former scenario are used as the initial iteration point of the next one. This greatly reduces the computation efforts in load flow solutions of the sequential scenarios. Simulation results show the effectiveness of the model and algorithm proposed.