Abstract:Fourier transform infrared（FT-IR） spectroscopy combined with chemometrics methods were used to discriminate different species of medicinal and edible plants of genus Dioscorea. The infrared spectra of five species of medicinal and edible plants of genus Dioscorea （Dioscorea opposita、Dioscorea bulbifera、Dioscorea henryi、Dioscorea hemsleyi、Dioscorea alata） were collected. The original spectra were optimized by 9 -point smoothing， multipoint baseline correction and automatic normalization. According to the preliminary analysis of spectra data， apart from D. hemsleyi， we could find that other four species of Dioscorea plants had high similar－ ities. The peaks at 1 154，1 081，1 021，928，763，577 cm-1 were strongest characterize the chemical composition of starch and some glycosides. The second derivative spectra in the region from 1 800 cm-1 to 400 cm-1 were se－ lected for hierarchical cluster analysis （HCA） and partial least square discriminant analysis （PLS-DA）. By HCA， 727 variables were extracted to build a rectangular array for obtaining the dendrogram. The rate of correct classification was 91.2 %. The results of PLS-DA showed that， the cumulative contribution of first six principal components was 97.3 %. At the same time， the classification result in scores plot was satisfactory. The study demonstrated that FT-IR spectroscopy coupled with HCA and PLS-DA could discriminate the different species of medicinal and edible plants of Dioscorea satisfactorily.