The dried stem bark of Berberis vernae C.K.Schneid., known as "Xiao-bo-pi" in Chinese, is a representative anti-diabetic herb in traditional Tibetan medical system. However, its anti-diabetic mechanisms and active components remain unclear. In this study, 1H NMR-based metabolomics, biochemistry assay, molecular docking, and network analysis were integrated to evaluate the anti-diabetic effects of B. vernae extract on type 2 diabetic rats, and to explore its active components and underlying mechanisms. Diabetes was induced by high-fat diet and streptozotocin. After 30 days of treatment, B. vernae extract significantly decreased the serum levels of fasting blood glucose, insulin, insulin resistance index, glycated serum protein, TNF-α, IL-1β, and IL-6, whereas significantly increased the serum levels of insulin sensitivity index in type 2 diabetic rats. A total of 28 endogenous metabolites were identified by 1H NMR-based metabolomics, of which 9 metabolites that were changed by diabetes were significantly reversed by B. vernae extract. The constructed compound-protein-metabolite-disease (CPMD) interaction network revealed the correlation between chemical constituents, target proteins, differential metabolites, and type 2 diabetes. Ferulic acid 4-O-β-D-glucopyranoside, bufotenidine, jatrorrhizine, and berberine showed good hit rates for both the 30 disease-related proteins and 14 differential metabolites-related proteins, indicating that these four compounds might be the active ingredients of B. vernae against type 2 diabetes. Moreover, pathway analysis revealed that the anti-diabetic mechanisms of B. vernae might be related to its regulation of several metabolic pathways (e.g., butanoate metabolism) and disease-related signal pathways (e.g., adipocytokine signaling pathway). In summary, B. vernae exerts a significant anti-diabetic effect and has potential as a drug candidate for the treatment of type 2 diabetes.