In this paper, a game theory based fuzzy QoS multicast routing scheme is proposed and has been implemented by simulation. It consists of three parts: edge evaluation, game analysis, and multicast tree construction. It does comprehensive evaluation on candidate edges based on adaptability membership degree functions for edge parameters, determines whether Nash equilibrium between network provider utility and user utility has been achieved on candidate edges by gaming analysis, and attempts to construct a multicast routing tree with not only user QoS requirements satisfied but also Pareto optimum under Nash equilibrium on network provider utility and user utility achieved or approached by the proposed algorithm. Simulation results have shown that the proposed scheme is efffective.