Abstract: In order to solve the problem of low tracking performance of the adaptive asymmetric joint diagonalization (AAJD) algorithm which is invalid in low signal to noise ratio (SNR), a high accuracy tracking algorithm for bistatic multiple-input multiple-output (MIMO) radar is proposed. Firstly, in order to solve the signal subspace expansion problem in the AAJD algorithm at low SNR, the eigenvalues are obtained by using the principal component order estimation principle. The steering vectors are sorted to obtain a more accurate signal subspace according to the size of the eigenvalues. Secondly, the multiple signal classification (MUSIC) algorithm is divided into two steps depending on the tracking status. The first step is to scan the whole airspace with a large step length, corresponding to tracking the unsteady state. The second step is to scan the smallspace with a small step length, corresponding to tracking the steady state. The range of airspace is determined by the angle of the previous time and the velocity of the target. The peak search process is changed to find the maximum value operation which further reduces the amount of calculation. The problem of the signal subspace expansion is solved in low SNR, improving the tracking performance. The algorithm uses improved MUSIC which has higher performance and lower calculation. Simulation results are presented to verify the efficiency of the proposed method.