Abstract: Realvalued domain processing has the advantage of reducing the heavy computational complexity for multipleinput multipleoutput （MIMO） radar angle estimation with large degrees of freedom. Unfortunately, realvalued domain processing cannot be applied to bistatic MIMO radar with the arbitrary array structure for direction of departure (DOD) and direction of arrival (DOA) estimation except additional preprocessing because the array do not have the characteristic of conjugate symmetry. Therefore, the multiple signal classification (MUSIC) super resolution algorithm for conventional array with the realvalued domain processing extends to the bistatic MIMO radar with the arbitrary array structure. According to the equivalent characteristic between conjugation and image of the steering vector of MIMO radar, the real part of the received signal covariance matrix is firstly extracted and an eigendecomposition is performed to obtain the signal subspace with “double targets” and its corresponding noise subspace. Then the characteristic of Kronecker product is used to reduce the dimension of the process to achieve one dimensional search MUSIC spectrum in the semirealvalued domain. The true DOD and its images estimations are substituted into reduceddimensional Capon spectrum to eliminate the peaks of images. Finally, the eigenvectors are utilized to obtain ambiguous DOA estimations and the minimum norm of the directioncosines difference method is used to disambiguate the DOA estimations. The proposed algorithm has the similar estimation performance and half computational complexity compared with the one dimensional search complex domain MUSIC algorithm. Moreover, the proposed algorithm can realize automatic pairing between DOAs and DODs. Simulation results verify the effectiveness of the proposed algorithm.