A Surface EMG signal classification method based on wavelet transform is presented in this paper. To utilize the nonstationary character of the EMG signals, dyadic wavelet transform is employed to obtain the signals' time-frequency representation. Singular value decomposition(SVD) is then used to extract feature vector for pattern classification. This motion classifier can successfully identify four types of forearm movement: hand grasp, hand extension, forearm pronation and forearm supination. Experimental result shows that this method has a great potential in the practical application of prothesis control.