Abstract: In the Gaussian noise background, an algorithm is proposed to jointly estimate the multiuser DOA and self-calibrate the mutual coupling error for Uniform Linear Array (ULA). First, the generalized spatial feature vector of each user is estimated by utilizing the Joint Approximative Diagonalization of Eigen (JADE) matrix method. Second a transformation matrix is defined, and based on which the generalized spatial feature vector is converted to the one which is related with part elements of the ULA. Then the multiuser coherent DOA estimates are obtained on the basis of the oblique projection and Forward and Backward Spatial Smoothing　(FBSS) methods. Finally, a mutual coupling self-calibration method is presented by utilizing the estimates of the DOA and the generalized spatial feature vector of each user. The computer simulation indicates that the algorithm has higher performance of DOA estimation accuracy and successful rate. The simulation results also demonstrate that, the proposed algorithm is universal for the situations where the mutual coupling error is known or not with white or colored additive Gaussian noise.