Abstract: In order to track the channel variation of MIMO-OFDM systems, traditional channel identification methods based on training sequence had to send large amount of redundant information, which results in lower efficiency of transmission. Meanwhile, the blind method based on subspace may have inevitable ambiguity and lower precision of identification. In this paper, an estimation method for the ambiguity matrix is proposed, which utilizes less training symbols to rectify the results from subspace method. The simulation results show that this proposed algorithm can reduce the amount of training symbols and increase the efficiency of transmission.