In this paper the MMAE(Multiple Model Adaptive Estimation) algorithm using the MGEKF(Modified Gain Extended Kalman Filter) of which modes are set to be measurement biases is proposed tu enhance the performance of target tracking with bearing only measurements. The state are composed of relative position, relative velocity and taregt acceleration. The mode probability is calculated from the bearing only measurements from the HMS(Hu 11-Mounted Sonar). The proposed algorithm is tested in a series of computer simulation runs.