A new approach for tracking maneuvering targets is presented. A target model that combines Singer's model with a deterministic step maneuver model is also proposed to account for the various realistic evasive maneuver strategies. A multiple model adaptive estimator incurs a mean tracking error in the presence of a pilot-com-manded abrupt target maneuver. A recursive real time least squares algorithm is used to estimate the acceleration of the target. This acceleration value is then fed to the multiple model algorithm for reducing the tracking error. A combined scheme of this input estimation filter and the mu1tiple model approach markedly improves the tracking accuracy. Simulation results confirm that the performance of the proposed algorithm is superior to that of the multiple model algorithm.