Abstract: For impulsive noise environments, a weighted sparse constraint on beam pattern is proposed to improve the limited interference suppression ability of the existing sparse constraint-based beamformers, which impose a same constraint on all signals from different directions. Firstly, a l1-norm sparse constraint is incorporated into the objective function of the minimum dispersion distortionless response beamformer by exploiting the sparsity of the beam pattern. Then, in order to force the covariance of the array received signal finite, the infinity-norm normalization algorithm is employed to adaptively pre-processed all snapshots received by the sensor array. Finally, a weighted matrix, which is constructed using the traditional subspace method based on eigen-decomposition of the obtained covariance matrix, is applied to sparse constraint term so as to enforce different constraints on signals coming from different angles. The experimental results show that the proposed beamformer not only keeps the relative low sidelobe level, but also deepens the null in the direction of the interferers significantly, which results in an improvement of the output SINR.