In this paper, by analyzing the parameters in CS-Jerk model and Kalman filter, a modified CS-Jerk model for target tracking algorithm was proposed. According to the residuals and its change rate, the fuzzy inference mechanism was used to adjust motor frequency of CS-Jerk model, and then, strong tracking filter in the motor model was used to compensate shortcomings of the Kalman filter. The simulation results show that the modified CS-Jerk model remarkably improves the tracking accuracy of high maneuvering target in different motor model in contrast with the original model, validating that the improved algorithm is reasonable and feasible.