For the problem that the traditional expert system method can not deal with the non-expecting situation, put forward the evolutionary expert system tree method to research on the air combat decision making problem. Taking middle range air combat as example, establish 2 dimension typical air combat model, combine the genetic method and expert system, establish the evolutionary expert system tree frame and tactics decision making system. Simulate and verify it under 1 VS 1 air combat environment. The simulation results show that the method can carry out decision making offline evolution under varied battlefield setting, the decision making system can adaptively make adjusting according to battlefield situation, and achieves the victory despite of a disadvantageous air speed setting in comparison with the opponent.