The output SINR of the existing MIMO radar beamforming algorithms may decrease greatly due to the large mismatch of the joint steering vector.To solve this problem, an algorithm using iterative robust Capon beamforming for MIMO radar is proposed.Considering the transceiver array structure of MIMO radar, the mismatch error model is analyzed in theory at first.Based on this, we point out the limitations of the big-uncertainty-set algorithm.Then, an algorithm using small uncertainty sets to constraint large mismatch is proposed.The proposed algorithm iterates over the estimated joint steering vector until meeting the designed termination conditions.Compared with the previous algorithms, the proposed method can acquire a more accurate estimation of the joint steering vector.Simulation experiments show that the proposed algorithm presents outstanding performance in improving the output SINR and has strong robustness in the case of large mismatch.