Abstract: In three dimension multiple-input multiple-output (3D MIMO) systems, large number of antennas are arranged at base station (BS) to obtain high spectral efficiency and large throughput. But increase of BS antennas also enhances the computational complexity of the multiuser precoding algorithm such as block diagonalization (BD) significantly. For the dimension increase of channel matrix, various ways can be chosen to obtain the equivalent matrix which can eliminate the user interference completely in the regular BD algorithm. Motivated by this observation, a low complexity precoding method based BD is proposed in this paper, which can simplify the procedure of the two singular value decomposition (SVD) operations in the regular BD algorithm prominently with the optimal equivalent matrix. Also with the 3D channel measurement data we validate the 3D capacity performance of the proposed method. Simulation results show that the proposed method can achieve the same capacity as the regular BD algorithm.