Abstarct: In this thesis we use adaptive estimators to improve the target tracking performance. In the target tracking by noisy measurements, target dynamics movement such as position, velocity and acceleration is estimated. Measurement equations and sometimes target motion equations are nonlinear, therefore to estimate the position and speed of target nonlinear estimation methods should be used. Here (In this work) we use extended, Unscented and Cubature Kalman Filters as estimator. Performance of Kalman Filters largely depends on the covariance of measurement noise and system noise that may be appeared by environmental conditions, modeling inaccuracies, errors and noisy sources. Several methods have been proposed to determine the covariance and one of these conventional methods are adaptive methods that have a proper performance. In this thesis, baxsed on innovation sequence and maximum a posterior proposed four adaptive approaches. Since the motion of a target has different models, one Kalman filter can’t be used for target tracking. To overcome this problem, the Interacting Multiple Model method that used in hybrid systems will be explained and the adaptation covariance will be discussed again. In the adaptive methods there is no guarantee for covariances to be positive definite. In this thesis a method for guarantee the positive definiteness of covariance is presented. To evaluate the adaptive methods in target tracking, with the introduction of bearing only target tracking assume two motion scenario and compare the adaptive methods baxsed on Kalman filters by using MATLAB such that the proposed methods improved target tracking performance in order to reduce the estimation error.