The quality of jasmine flowers in terms of flavour, medicinal and nutritional uses is influenced by the factors of its origin. Hence, the origin traceability of jasmine flowers is of great significance in protecting the rights and interests of consumers and promoting the healthy development of the jasmine industry. In order to discriminate the geographical origin of Jasmine, a hundred Jasmine samples from four main producing districts, including Hengzhou of Guangxi, Qianwei of Sichuan, Fuzhou of Fujian and Yuanjiang of Yunnan, were collected. Near-infrared spectra, (900~1 700 nm) of those samples were acquired using integrating sphere and fibre-optics probes. Savitzky-Golay (SG) spectral smoothing and multivariate scatter Correction (MSC) were used for spectral pre-processing. After the spectral pre-processing, a jasmine origin discriminant model was developed using PCA combined with linear discriminant analysis (LDA) and k-nearest neighbor (KNN). In the modelling process, 68 samples were used as the training set and 32 samples were used as the test set, and the model parameters were optimised by interaction tests. The results show that the discriminant models based on both PCA-LDA and PCA-KNN have good prediction ability, in which the prediction accuracy of both methods reaches 100% for the spectral data obtained by integrating sphere sampling, and the prediction accuracy of PCA-LDA and PCA-KNN for the spectral data obtained by fiber optic probe sampling is 100% and 93.75% respectively. Finally, a comparative analysis of the chromatographic fingerprint profiles of jasmine flowers from different origins further elucidated the material basis for identifying jasmine origins based on NIR spectroscopy. Thus, this work provides a fast, environmentally friendly, and accurate method to trace the geographical origin of Jasmine, which is meant for the protection of the place of origin for Jasmine.