Abstract: An improved sorted QR decomposition(ISQRD) MIMO detection algorithm is proposed and its performance is analyzed.When the sorted QR decomposition detection(SQRD) algorithm is used in detection,high bit error rate(BER) may occur.In order to solve this problem,the ISQRD algorithm transforms the channel matrix column orthogonally to avoid solving the upper triangular matrix,and sorts the channel matrix from small to big by 2-norm just one time.In the process of detection,a parallel processing method is used to feedback a part of the judged signals and to eliminate interferences in the received signals,and the detecting performance is improved distinctly.Simulation is made in the wireless communication environment where lots of scattering objects exist,and the results show that computation complexity of the proposed algorithm is reduced and the detection performance is improved dramatically in comparison with the traditional sorted QR decomposition algorithm.