To solve the problem of Kalman filter not applicable in nonlinear system for bearings-only target tracking, the Extended Kalman Filter (EKF) algorithm is presented. EKF algorithm make first-order linearization truncated to Taylor expansion of nonlinear function and ignore the rest higher-order items, which is to make the nonlinear problem into linear, thus the Kalman filter applicable in nonlinear system. In simulation, we simulate the EKF algorithm in Constant Acceleration (CA) model. The results indicate that the EKF algorithm convergence and has favorable target tracking performance applicable to target tracking system .