Abstract: The insufficient number of space-time snapshots collected by the Multiple-Input Multiple-Output (MIMO) sonar in the underwater target detection scene results in the performance degradation of Space-Time Adaptive Detection (STAD). To deal with this problem, a robust STAD method for the MIMO sonar is proposed. Firstly, the persymmetric structure is extended to the MIMO sonar and the number of space-time snapshots is expanded to twice the original through this structure; secondly, iterative adaptive approach is utilized to reconstruct the space-time covariance matrix; finally, the test statistic for the MIMO sonar is obtained based on the Generalized Likelihood Ratio Test (GLRT) criterion. Computer simulation results show that the proposed method can realize the accurate estimation of space-time covariance matrix, greatly reduce the demand of snapshots in STAD, and improve the robustness of target detection.