In this paper, demodulation performance of first-order cyclic statistics and second-order cyclic statistics for amplitude modulation signals is introduced. By theoretical research, it is proved that cyclic mean does not possess demodulation performance for amplitude modulation signals, but cyclic autocorrelation function can extract frequency components from amplitude modulation signals. Based on the above, both cyclic autocorrelation function and envelope demodulation based on Hilbert transform are compared. The results show that the two kinds of analysis methods have similar performance for demodulation, in the mean time it is pointed out that they also possess essential distinction, and cyclic autocorrelation function is better to demodulate amplitude modulation signals. In this paper, demodulation performance of first-order cyclic statistics and second-order cyclic statistics for amplitude modulation signals is introduced. By theoretical research, it is proved that cyclic mean does not have demodulation performance for amplitude modulation signals, but cyclic autocorrelation function can extract frequency components from amplitude modulation signals. Based on the above, both cyclic autocorrelation function and envelope demodulation based on Hilbert transform are compared. The results show that the two kinds of analysis methods have similar performance for demodulation, in the mean time it is pointed out that they also possess essential distinction, and cyclic autocorrelation function is better to demodulate amplitude modulation signals.