Based on decision theory algorithm，the characteristic parameter extraction and automatic identification technology of mixed modulation signal are researched,and then the tree classifier with identification steps which are suitable for mixed modulation signal modulation recognition are put forward. The characteristic vector which are composition of the number of subcarrier signal, envelope variance of mean normalization and algorithm of the statistical value of subcarrier signal instantaneous amplitude distribution area are first used in recognition of outer modulation and inner modulation respectively so as to reduce the noise interference and improve the accuracy of characteristic parameters. The simulation results show that modulation recognition rate is close to 90% under the condition of SNR which is 6 dB and it has a good recognition effect compared with mixed modulation recognition method existing and a broad prospect of application in the management of the mixed signal modulation identification.