Abstract: The objective of this paper was to analyze the difference of composition between hybrid sorghum and conventional sorghum in southwest China, and to establish the method of variety identification. The basic physical and chemical components of 29 sorghum samples were determined, and the metabolite composition of 29 sorghum samples was determined by ultra high performance liquid chromatography coupled with Q Exactive HF-X mass spectrometry. The marker component screening and construction of variety identification model were carried out using stoichiometry, which included principal component analysis (PCA), orthogonal partial least squares discriminant analysis (OPLS-DA), cluster analysis and receiver operating characteristic (ROC) curve analysis. The model was based on the basic physicochemical composition and all metabolites. The results showed that the contents of amylopectin and tannin of the conventional varieties form southwest China were higher than that of hybrid varieties, while the contents of amylose and 1000-grain weight were lower. By analyzing the physicochemical properties, it was possible to identify certain samples. The OPLS-DA model was constructed with all 1048 compounds variables, and 46 of the differential metabolites that had a significant impact on the model identification were screened out. A novel OPLS-DA identification model was developed for sorghum varieties, achieving a 100% accuracy in identification. The conventional varieties from southwest China exhibited a relatively high concentration of phenolic compounds. Furthermore, after conducting receiver operating characteristic (ROC) curve analysis and assessing the relative standard deviation (RSD) of quality control sample (QC), 20 feature components that exhibited both high diagnostic accuracy and good reproducibility were identified. Among them, syringetin 3-glucoside and 6"-o-acetyldaidzein had higher expression abundance in conventional sorghum varieties, while D-glucarate, putsutrine, N-acetylputsutrine and L-glutamine had higher content in hybrid sorghum varieties. This study unveiled the disparity in metabolites between hybrid sorghum and conventional sorghum. The study successfully developed a scientifically rigorous and precise untargeted metabolomics approach using UHPLC-Q Exactive HF-X-MS. This method can be effectively utilized for the identification of sorghum varieties, offering a novel strategy for sorghum variety identification.