A surface-enhanced Raman spectroscopy（SERS）with chemometrics was developed for the rapid detection of methotrexate and voriconazole in mixtures.An gold nanocomposite(AuNPS-PDDA) subtrate was fabricated to obtain the SERS spectra signals of methotrexate and voriconazole,which were transformed into wavelet space by continuous wavelet transform（CWT）,then analyzed by features matching in wavelet space（FMWS）.This method could alleviate the influence of baseline variations and random noise,with detection limits for methotrexate and voriconazole of 1.0×10-8 mol/L and 2.86×10-4 mol/L,respectively.The feature matching coefficients for methotrexate and voriconazole were both higher than 0.96,which were significantly better than the traditional hit quality index.Besides,the actual ratio for two kinds of substance in the mixtures was determined by nonnegative least squares method,and the correlation between predicted values and real values was larger than 0.98.The results confirmed that the proposed method could be applied in the effective identification and semi-quantification of target compound in mixtures.