Growing numbers of web services that offer identical functionality but differ in non-functional properties are emerging on the network, to the need to select them to form a composite service to meet user’s requirements has become one research hotspot. Web service selection methods are an attempt to to find optimal solutions for users. However, because each user’s personal preference is different and web services are massive and dynamic, it is hard to find optimal solution. Therefore, a Hybrid Algorithm combining Ant Colony Algorithm and Genetic Algorithm for web service composition is proposed in this paper. The global optimization problem in web service composition is firstly transformed to the problem of finding an optimal path in the weighted directed acyclic graph with certain QoS (Quality of Service) constrains. And then an improved ant colony algorithm and an improved genetic algorithm are used alternately in the hybrid algorithm. Improved ant colony algorithm is used to achieve the non-dominated solution sets. Using the sets as the initial population sets, improved genetic algorithm is performed to assist ant colony algorithm to obtain the optimal solution. Experimental results demonstrate the validity and efficiency of the proposed algorithm.