Abstract: To improve the blind equalization performance for complex noncircular signals, the widely linear (WL) filtering theory is deeply analyzed. The paper takes advantage of simplicity and robustness of the constant modulus (CM) criterion to construct the reduced complexity widely linear blind equalizer. A reduced complexity widely linear recursive least mean square CM blind equalization algorithm is produced. The new algorithm directly uses the real and imaginary parts of received signals as input of the equalizer, thus obtains the full second-order information of the real and imaginary parts. The proposed scheme takes the real operation instead of complex operation and obtains the computational complexity which is just a few more than the traditional linear blind equalization algorithm and much less than the standard WL filter. Simulation results show that the new algorithm can simultaneously decrease the residual inter-symbol interference (ISI) and symbol error rate when compared with the traditional blind equalization algorithms.