This paper presents a new on-off beamforming algorithm to provide an optimal weight vector. The proposed algorithm is based on the eigen-space on-off algorithm for exploiting a diversity gain that is available in the signal environments of wide angle spread. The proposed technique utilizes the primary and secondary eigenvectors to form the optimal weight vector, which is robust to the angle spread. It is confirmed in this paper that the proposed algorithm shows excellent performance compared to some typical beamforming methods which degrade the receiving performance as the angle spread increases. It is also shown that the proposed algorithm has a reasonable computational complexity of O(14N) and fast convergent speed, where N is the number of antenna elements.