Abstract: A computationally efficient direction-of-arrival (DOA) estimation algorithm for noncircular signals based on multistage wiener filter theory and polynomial rooting technique is proposed. Firstly, the array extension matrix is constructed by utilizing the noncircularity of the signals. Secondly, the signal subspace is achieved with multistage Wiener filter (MSWF), which does not require the formation of the covariance matrix and its eigendecomposition. Thirdly, the polynomial rooting method of the signal subspace is derived based on uniform linear arrays, by which the estimate of DOA are obtained. The proposed algorithm is more computationally efficient, especially for the case of large number of sensors, in comparison with NC-root-MUSIC, NC-ESPRIT and extended propagator method (EPM), although their performances are near the same.