Abstract: Space recovery space-time adaptive processing method based on sparse recovery (SR-STAP) improves the performance of moving target detection. However, when the off-grid effect occurs, the performance of SR-STAP algorithm was degraded. Hence, an off-grid error iterative self-calibrated STAP algorithm was proposed to solve this problem. Firstly, the grid points with high power spectrum value were selected from the general global dictionary to construct the large dictionary. Secondly, the atoms with high matching degree with clutter points were found from the large dictionary as the center, and the local dictionary was constructed. In addition, the local search was carried out by considering the posterior Bayesian, and the highest matching degree with clutter points was founded. Finally, the calibrated dictionary and the optimal STAP filter weight are obtained, which optimizes the performance of STAP algorithm in off-grid situation. The effectiveness of the algorithm is verified by simulations.