The FTIR data collected from 36 normal SD rats, 60 1,2-DMH- induced SD rats, and 44 second generation rats of those induced rats were first preprocessed. Continuous wavelet transform was used in the analysis of FTIR. The detail signal, L1, is almost all the high frequency signal. Difference between a few adjacent detail signals are little. L5, L10 and L15 are sensitive to the change of the absorption frequency in the spectra. They have strong response to three feature peaks of the original signal. So the detail signals of the three scales were used as feature vector space. 12 feature variants were extracted using continuous wavelet analysis. Based on BPNN classification, all spectra were classified into two categoriesnormal and abnormal ones. The accuracy values of identifying normal, dysplastic, early carcinoma, and advanced carcinoma were 100%, 95%, 97.5%, and 100%, respectively.