An adaptive model of rotor/turbo-shaft engine on board was constructed using Kalman filter. The state variable model of rotor/turbo-shaft engine was built through the data fitness method, and a Kalman filter was designed using the deterioration factors as augmented state variables. The Kalman filter can be utilized to predict the component deteriorations from the deviation of measurable parameters, which are used to trim the unmeasurable output parameters of the model on board so that it can adapt the actual engine. The digital simulation results show that the adaptive model on board is able to match the actual engine well. Meantime, the robustness and real-time of the adaptive model are also validated in the full flight envelope.