Abstract: A new algorithm based on spectral correlation of the square of MPSK signals is proposed for recognition of MPSK .Firstly the theory of cyclic spectral is explained and analyzed by the paper, a class of new unitary characteristic parameters is proposed for the modulation recognition of MPSK signals from the expression of cyclic spectrum function, which is invariant with local carrier phase offset and amplitude scale of the base band MPSK signal. The method of feature extraction is very simple and efficient, and it is conducive to project implementation. Finally the algorithm is simulated by computer. Include the simulate of recognition performance and the simulate of performance comparison with similar algorithms. The good performance in lower signal noise ratio(SNR) of the algorithm has been proved by the simulation results. In the gauss white nose channel(AWGN), simulation results show that the success rate of recognition is over 95%when the SNR is 4dB.Theoretical analysis and computer simulation prove the robustness and efficiency of the new recognition algorithm.