To successfully shoot down a flying enemy missile, high accuracy missile tracking is a pre-requisite. However, since the dynamics of the missile change with the phase of the trajectory, there is a limit to the tracking with ordinary Kalman filter (KF) which considers single dynamics. Therefore, in this paper, to improve tracking performance, interacting multiple model (IMM) which considers various dynamic models simultaneously is applied. In addition, by classifying the missile head and debris generated at stage separation using the model probability calculated in the IMM, unnecessary detection and tracking resources used to track debris are reduced. The performance of the proposed method is verified through various simulation results and the favorable results are obtained.