In service-oriented computing, negotiation can help service provider and consumer reaching an agreement between services. This paper discusses the optimization of utility in service negotiation. In one-toone negotiation with multiple issues, two parties can reach the final equilibrium point using the concession negotiation strategies. Using traditional concession strategies, each party cares only about the maximization of its own utility rather than of the joint utility. In many applications, this does not ensure each party can reach the maximum point of utility. We propose an improved concession strategy in this paper, which allows further optimization after reaching the preliminary results of negotiation, so that the joint utility between two parties can be improved. This strategy can be used in the applications that need dynamic finding, evaluating, selecting, and binding of services, which helps service provider and consumer reaching an agreement. The results of simulation show that the strategy of joint utility optimization is effective.