Ant colony algorithm is one kind of new heuristic biological modelling method which has the ability of parallel processing and global searching, but its convergence speed is slow because of poor pheromone on the early path. In this paper, discuss a new algorithm which combines genetic algorithm and Ant colony algorithm. Genetic algorithm is added to ant colony algorithm's every generation in the proposed algorithm. Making use of genetic algorithm's advantage of whole quick convergence, ant colony algorithm's convergence speed is quickened. Genetic algorithm's mutation mechanism improves the ability of ant colony algorithm to avoid being trapped in a local optimal. The simulation shows that the new algorithm is effective in solving distribution network planning problem.