In recent years, wireless network coding has attracted a significant interest from the research community. Wireless network coding improves the network performance, e.g., increasing throughput and minimizing delays, by taking advantage of the side information obtained by the nodes. We focus on the scenarios in which the clients are selfish, i.e., their goal is to maximize their own utility, rather than the utility of the system. Our objective is to motivate selfish clients to contribute to the information exchange. For both server-client and point-to-point communications, we propose the incentive-compatible network coding mechanisms that achieve the social optimum. Since some of the related problems are intractable, we present approximation algorithms with provable performance guarantees.