Abstract: In this paper, based on symmetric uniform linear array (ULA), a direction-of-arrival (DOA) estimation algorithm was proposed for incoherently mixed point sources and distributed sources. The spatial differencing technique was adopted to separate the two kinds of sources by exploiting the difference of covariance matrix structure between point sources and distributed sources. For the point sources, the traditional MUSIC algorithm was used to estimate the DOA. For the distributed ones, the DOA was estimated by exploiting the rotational invariance of the signal subspace. Our algorithm not only eliminated the effects of the point sources on the distributed ones, but also did not need to estimate the distribution parameters, which decreased the computational complexity greatly. With a symmetric ULA of 2N+1 elements, 2N mixed sources can be estimated, including N distributed sources at most, which effectively decreases the array aperture loss. Simulation results show that the proposed algorithm outperforms the generalized eigenvalue decomposition (GEVD) algorithm.