Correct identification of the origins of herbal medical products is becoming increasingly important in tandem with the growing interest in alternative medicine. However, visual inspection of raw material is still the most widely used method, and newer scientific approaches are needed. To develop a more objective and efficient tool for discriminating herbal origins, particularly Korean and Chinese, we employed a nuclear magnetic resonance (NMR)-based metabolomics approach combined with an orthogonal projections to latent structure-discriminant analysis (OPLS-DA) multivariate analysis. We first analyzed the constituent metabolites of Scutellaria baicalensis through NMR studies. Subsequent holistic data analysis with OPLS-DA yielded a statistical model that could cleanly discriminate between the sample groups even in the presence of large structured noise. An analysis of the statistical total correlation spectroscopy (STOCSY) spectrum identified citric acid and arginine as the key discriminating metabolites for Korean and Chinese samples. As a validation of the discrimination model, we performed blind prediction tests of sample origins using an external test set. Our model correctly predicted the origins of all of the 11 test samples, demonstrating its robustness. We tested the wider applicability of the developed method with three additional herbal medicines from Korea and China and obtained very high prediction accuracy. The solid discriminatory power and statistical validity of our method suggest its general applicability for determining the origins of herbal medicines.