The blind source separation (BSS) was dealed with in the overdetermined case, i.e. the number of receive an-tennas was larger than that of sources. Firstly, a method which converted the overdetermined BS into complete BSS was given. Secondly, based on the relation between the optimal solution of generalized Rayleigh quotient and the convergence solution of fourth-order cumulant cost function, a maximum ratio combining (MRC) aided BSS algorithm (MRC-FastICA) was proposed. With this method, FastICA was used to obtain multiple estimated ls corresponding to a same source, and then these estimated signals were combined by our MRC method. Simulation results show that, compared with tradi-tional algorithms, the proposed algorithm can obtain diversity gain, and achieve better SINR performance.