Rhizomes of the Polygonatum species are well-known in traditional Chinese medicine. The 2020 edition of Chinese Pharmacopoeia includes three different species that possess different pharmacological effects. Due to the lack of standardized discriminant compounds there has often been inadvertently incorrect prescriptions given for these medicines, resulting in serious consequences. Therefore, it is critical to accurately distinguish these herbal Polygonatum species. For this study, UPLC-Q-TOF-MS/MS based metabolomics was employed for the first time to discriminate between three Polygonatum species. Partial least squares discriminant analysis (PLS-DA) models were utilized to select the potential candidate discriminant compounds, after which MS/MS fragmentation patterns were used to identify them. Meanwhile, metabolic correlations were identified using the R language package corrplot, and the distribution of various metabolites was analyzed by box plot and the Z-score graph. As a result, we found that adenosine, sucrose, and pyroglutamic acid were suitable for the identification of different Polygonatum species. In conclusion, this study articulates how various herbal Polygonatum species might be more accurately and efficiently distinguished.