Abstract: Considering the strict preprocessing requirement of modulation recognition algorithm in fading channels, an algorithm of modulation classification based on cyclostationary test is proposed to promote the classification performance under blind reception conditions. Combing cyclostationary test, the signal cyclic frequency features of cyclic cumulants are used to discriminate common digital modulation signals, such as OFDM, PSK, FSK, etc.. The method obviates the need for preprocessing tasks, such as estimation of parameters and synchronization. The theoretical derivations and simulations simultaneously confirm the effectiveness of discriminating signals in the fading channel with low SNR. Furthermore, this method can be used to promote the signal blind classification performance in fading channels under the non-cooperative reception conditions.