Abstract: Abstract The traditional adaptive robust beamformer is sensitive to various of array imperfections. And the output array performance suffers deterioration, especially in the presence of large direction of arrival (DOA) mismatch. To overcome performance degradation, a new iterative robust beamformer is introduced with adjustable error radiuses. Each radius is the optimal solution of a quadratically constrained quadratic program (QCQP) problem, which is derived from based on the subspace principle. Additionally, the output signal to noise and interference (SINR) is improved on the fact that all the estimated mismatch levels are smaller than the ones used in the conventional methods. Moreover, the proposed method is extended to tackle with multiple mismatches together using adjustable ellipsoid uncertainty sets. Finally, the new beamformer is compared with other approaches in the simulations. And its superiority is demonstrated.