Doppler shift and signal power attenuation in the complex environment both can make deviation bigger or even lose lock. Because the conventional Kalman filter can’t meet the dual requirements of high dynamic and high sensitivity when tracking maneuvering targets, a new carrier tracking method based on variable dimension Kalman filter was presented. By introducing motor model and non-monitor model of the carrier tracking and through putting in a maneuvering detector to monitor target movements, the real-time and effective switching of carrier tracking model was realized, so the adaptive tracking of carrier in motor and non-monitor state was achieved and the error caused by the change of maneuverability was overcome. 〖JP3〗Theoretical analysis and simulation results show that: comparing with the tracking algorithm based on standard Kalman filter, the phase tracking error of this algorithm is reduced by about 37.5% and the frequency tracking error is reduced by about 77% in weak carriertonoise ratio as low as 30dBHz environment.