Abstract: Based on the study of cyclostationarity and nonlinear transform of multilevel CPM signals, in this paper a novel blind identification algorithm is proposed to estimate modulation order. According to the theory of cyclostationary,some discrete spectral lines,located in the linear combination of communication signals' carrier frequencies and symbol rates,are present in nonlinear transform of the signals. Some basic parameters can be estimated by extracting the spectral lines of the nonlinear transform. After the analysis and derivation of a simplified expression for the first order cyclic moment of CPM signals, the spectral lines feature related with modulation order is investigated, followed by the algorithm steps. The experimental results sustain our theoretical analysis and indicate that the proposed algorithm with lower computational complexity, is independent of symbol period and insensitive to noise, compared with the existing algorithm. This method can be used to accomplish effective parameters estimation of single-h CPM signals with any Frequency-Shaping Pulse in the AWGN channels.