Abstract: It is important research contents to determine the direction-of-arrival (DOA) of radiation sources in the array signal processing, which has been widely used in radar, sonar and wireless communication system. This paper studies the (DOA) estimation problem of far field narrowband signal sources. We build up the sparse representation model of far field narrowband using the spatial sparse property of the DOA of the incident sources. We propose a DOA estimation method combined the eigenvalue decomposition and multiple orthogonal matching pursuit algorithm (MOMP) based on the eigenvalue decomposition of array output covariance matrix and the theory of greedy matching pursuit algorithm. Firstly, we use eigenvalue decomposition to reduce the dimension of the received data form array output. The reduction operation transforms the DOA estimation problem into solving a multiple measurement vectors (MMV) problem which is appropriate for the MOMP algorithm. Then, the MOMP algorithm is employed to estimate the DOA from the data of reduced dimension. High resolution DOA estimation is achieved in low signal to noise ratio (SNR) and less computational complexity is guaranteed. The performance is compared with the representative DOA estimation methods including MUSIC, Capon and OMP. Simulation experiments are conducted to validate the effectiveness of the proposed method regardless of the coherence of the incident signals.