From the point of view of the improved population size and genetic operator, a new genetic algorithm with adaptive population size is proposed in order to solve distribution network recovery reconfiguration problem. It gives the individuals in the group with age and life-time instead of selection operator and uses the strategy based on loop to reduce the generation of infeasible solution. Macro-control is used to solve the expansion problem of the population size. This paper adopts self-adapting crossover and mutation rate to improve the efficiency of the algorithm from the global and local perspective. It is proved an effective and stable algorithm according to an example in the distribution network recovery reconfiguration.