Aimed at the problems that the maneuvering targets are restricted by some known physical constraints, a multiple model (MM) method is adopted. In accordance with various motion behavior of maneuvering targets, a composed multimodel method is used to approximate. And in the light of the known physical constrains of targets, a moving horizon estimation (MHE) is used to process, and a priori information of state estimation is utilized to promote the precision of estimation. To incorporate the MM method into MHE framework, through an estimation evolution formula and the modified update formula for the estimation covariance matrix, a MMMHE optimization and algorithm are finally nally presented for the tracking problem. The simulation result shows that compared with the adaptive Kalman filter lter (AKF) and the interacting multiple model (IMM) method, this algorithm can perform well a task of tracking maneuvering targets (especially for the physically constrained motion condition)