The geographical discriminations of 104 Longjing tea samples collected from three tea-producing areas （Xihulongjing tea-producing area， Qiantanglongjing tea-producing area and Yuezhoulongjing tea-producing area） were studied based on HPLC fingerprints combined with stoichiometric determination analysis. The characteristic chromatograms of Longjing teas were constructed by HPLC-DAD detection， which showed that the reference fingerprints were composed of 28 characteristic common peaks including 16 known components and 12 unknown components. The geographical-recognition accuracies of Longjing teas exceeded 90% by Discriminant Factor Analysis（DFA）， Multilayer Perceptron （MLP） and Support Vector Machine （SVM）， especially by SVM with the accuracy of 98.08%. In conclusion， this study indicated it was feasible for geographical discrimination by multiple HPLC fingerprints combining with chemometrics methods.