Abstract: The direction-of-arrival (DOA) can be estimated as measured the distance between each search steering vector and the noise subspace or signal subspace with MUSIC algorithm. Therefore, the subspace deviation which associated with the correlation matrix will deteriorate the performance of MUSIC algorithm. To alleviate this decreasing in DOA estimation with secondary data deficient scenario and/or strong and weak signal coexistence, a new method based on pseudo-noise subspace projection is presented. The approach is performed in two stages. First, we employ a modified correlation matrix at each search steering vector to calculate the pseudo-noise subspace, then, the spatial spectrum can be obtained as weighted the MUSIC spectrum with the projection value of the search steering vector on the corresponding pseudo-noise subspace. The high-resolution of subspace processing is remained and the robustness against small sample support and in the presence of strong signal and weak signal is improved. Theoretical analysis and numerical simulation indicate that its performance is better than that of MUSIC.