The problems of reactive power optimization with genetic algorithm are discussed in detail in this paper.The crossover,mutation and inversion operations are proposed which not only reduces the search space,but also avoids the infeasible solutions produced during initialization and gene operations.The proposed genetic algorithm has been tested in an IEEE 30-bus power system.At the same time,based on the above genetic algorithm,network less of electric power systems can be effectively reduced,and then reactive power optimization can also be realized.