Abstract: Compared with a single-sensor system, a multi-sensor system can obtain the information of the detected target to a great extent. However, data detected by unmanned aerial vehicle sensors in air combat is deceptive to a certain extent. Using the game fusion technology to fuse multi-sensor data can obtain information in line with the actual needs of air combat. A multi-source air combat information game fusion technology is studied based on the improved Dempster-Shafer (D-S) evidence theory. Before information fusion, Jousselme distance is employed for preprocessing, and Fisher information is used for a conflict data game to obtain a strategy set to make air combat data reliable. On this basis, the fusion method based on D-S evidence theory is improved according to the Deng entropy method. Compared with the traditional D-S evidence theory method, the fusion data conform to those in actual air combat. Finally, a game fusion simulation is performed on conflicting multi-source air combat information. The simulation results verify the feasibility and advantages of the developed method.