Non-volatile compounds play a key role in the quality and price of Keemun Black Tea (KBT). Normally, the non-volatile compounds in KBT samples from different producing areas vary greatly. The development of rapid methods for traceability of KBT geographical origin is meaningful. In this study, we develop KBT geographical origin discrimination models based on non-volatile compounds.Seventy-two KBT samples were collected from five towns in Anhui Province to determine 13 KBT compounds by high-performance liquid chromatography (HPLC). Analysis of variance showed that the content of 13 compounds in KBT indicated significant differences (P<0.05) among five towns. Three multivariate statistical models including principal component analysis (PCA), soft independent modeling of class analogy (SIMCA), and linear discriminant analysis (LDA) were built to origin discriminant. PCA effectively extracted three principal components namely theaflavins, galloylated catechins, and simple catechins. Through SIMCA, high sensitivity (64.5-99.2%) of class model was achieved. Furthermore, to establish the discriminant functions, six variables (gallic acid, (+)-catechin, (-)-epigallocatechin gallate, theaflavin-3-gallate, theaflavin-3,3'-di-gallate, and total theaflavins) from 13 variables were elected, and LDA was applied, which performed satisfying overall correct classification rate (94.4%) and cross-validation rate (88.9%) of KBT samples.The results showed that HPLC analysis together with chemometrics is a reliable approach for tracing KBT and guaranteeing their authenticity. This article is protected by copyright. All rights reserved.