Abstract: On the basis of continuous wavelet transform, a signal analysis method based on extracting the energy spectrum of the signal on wavelet scale was proposed. This method was then applied to analyzing the gear vibration signal of different wearing conditions. The result shows that the energy spectrum of the signal on scale is closely related to the wearing condition of the gear. The accumulation values of the energy spectrum of the signal on scale are computed; then a curve much like that of the gear wear course is fitted according to these values. The curve shows that the energy spectrum of the continuous wavelet transform of the signal is sufficient to represent the wearing condition of the gear. A feature vector extraction method based on continuous wavelet transform is represented and applied in extracting the wearing feature of the gear. The distances between the vectors of different wearing condition proved that this feature extraction method is quite effective.