In a real peer-to-peer (P2P) network, large amounts of network measurement results show that free riding is prevalent in almost all P2P reliable streaming media networks, which reduces the robustness, availability, service response speed, and lifetime of P2P reliable streaming media networks. Research of the reasonable and effective mechanisms to prohibit free-riders and incite selfish nodes to contribute more to the system has become an important direction for application research of P2P reliable streaming media network. Analysis by the intrinsic characteristics of free-riding and the related impacts on system performance, the behaviors of P2P reliable streaming media nodes are modeled and an idea with the goal of keeping moderate safety by allowing some errors but no crimes is introduced without sacrificing overall performance. Furthermore, the game theory is used to restrain free-riders and encourage them to be more altruistic. Reputation, contribution, and revenue of each node are adopted as metrics to assess the model. And the existence of Nash equilibrium for the model is proved; the rules, constraints, and a detailed analysis of it are given as well. Simulations show that the proposed model is effective in countering free-riding behavior, improving the performance and quality of service (QoS) of the P2P reliable streaming media network. It is able to keep relative balance.