This paper proposes a metaverse sensing data offloading method based on game theory. Considering the benefits and costs of data offloading for virtual service providers, a metaverse data offloading model is built and the utility functions of service providers and edge server are designed. Then, the interaction process between service providers and edge server is modeled as a non-cooperative game model. The maximum optimization problem is formulated by taking into account computing latency and energy consumption. It is theoretically proved that there is a unique Nash equilibrium solution for the game process between service providers and edge server. Simulation results show that the proposed metaverse data offloading method can efficiently utilize system resources and maximize the utilities of edge server and service providers.