Gentiana rigescens ("Dian Longdan" in Chinese) medicinal plant is usually used for its activities of liver protection, cholagogic, anti-inflammatory, anti-fungal, anti-hyperthyroidism, anti-hypertension, hyperglycemia, and relieving spasm and pain. In this study, methods for the discrimination of different geographical origins of G. rigescens by FTIR spectroscopy in hyphenation with chemometric methods were developed. Different pretreatments including standard normal variate, multiplicative scatter correction, first or second derivative, Savitzky-Golay filter, and Norris derivative filter were applied on the spectra to optimize the calibrations. According to spectrum SD, spectrum ranges (3559-2709 and 2026-756 cm(-1)) were selected, and principal component analysis-Mahalanobis distance (PCA-MD) model was built [the cumulative contribution rate of the first 10 principal components, determination coefficient (R2), root-mean-square error of calibration (RMSEC), and root-mean-square error of prediction (RMSEP), and prediction accuracy were 96.4%, 98.6%, 0.5031, 0.7758, and 96.23%, respectively]. The spectral regions (3791-3442, 3043-2765, and 2013-646 cm(-1)) were selected by using the variable importance in projection, and partial least squares discriminant analysis (PLS-DA) model was built (the cumulative contribution rate of the first 10 principal components, R2, RMSEC, RMSEP, and prediction accuracy were 91.3%, 92.0%, 0.1171, 0.1806, and 100%, respectively). This research showed that FTIR spectroscopy in combination with chemometrics methods (PCA-MD and PLS-DA) was suitable for the discrimination of different geographical origins of G. rigescens. Furthermore, it was found that PLS-DA provided better results than PCA-MD.