With the growing number of alternative Web services in the open Web service environment, users have put forward new demands that bind the runtime of Web services, which requires as much short computation time as possible to satisfy user’s end-to-end QoS requirements service composition. Therefore, this paper proposes a Web service dynamic selection approach, based on the decomposition of global QoS constraints, WSDSA (Web service dynamic selection approach). The WSDSA uses as adaptive adjustment method (AAM), based on fuzzy logic and adaptive particle swarm optimization algorithm (APSO), to adaptively decompose global QoS constrains to local constraints with the user’s preferences, and then WSDSA can obtain the most appropriate composition service with local selection. Performance evaluations show WSDSA is very effective, and is able to reach optimal or near optimal results with a very low time cost, which satisfies the real time and dynamic service selection.