A new digital modulation recognition method based on instantaneous information is proposed for the application under low signal-to-noise ratio（SNR）.An improved wavelet threshold de-noising algorithm is used for instantaneous information de-noising

which can improve recognition ability under low SNR.The method can identify and classify seven digital signals which are MASK

MFSK

MPSK and MQAM and so on

by using BP neural network with resilient backpropagation（RPROP） training algorithm as the classifier.The computer simulations show that the proposed algorithm effectively improves the practicability because of an overall success rate of 96 % at the SNR of 2 dB.