The most a vailable direction of arrival (DOA) estimation algorithms require covariance mat rix estimation and eigendecomposition, or even its inversion, thus increasing the computational complexity. Here a novel sub aperture multiple sigal classification (MUSIC) algorithm f or DOA estimation based on fast Fourier transform (FFT) is proposed. Firstly, ea ch received data vector of uniform linear array (ULA) is portioned into four sub vectors. Then FFT is applied to each sub vector to achieve the coherent i n tegration. By utilizing the data corresponding to the peaks of coherent integrat ion in each sub vector, a reduced dimensional data vector is constructed for D OA estimation in terms of MUSIC. Since the full dimensional covariance matrix e stimation and eigendecomposition are avoided, the computational complexity is re latively low. Numerical examples are provided to verify the effectiveness and su periority of the proposed method.