Agro-herbal materials vary in prices and qualities depending on the origin and age and the differentiation is both scientific and public health issue. Here, we describe a metabolomics approach used to discriminate ginseng roots from different sources. Six different types of ginseng roots from China and Korea were analyzed by NMR-based metabolomics. Chinese Dangsam showed prominent differences and was easily differentiated. The difference was mainly due to the large signals in the sugar region. We further analyzed the metabolomics results in subgroups. Jeonra (Korean), Choongcheong (Korean), and Chinese ginseng in subgroup 1 could be easily differentiated by the first two principal components. The loading plot for PC1 showed that the Jeonra and Chinese ginseng roots were mainly separated by sugar signals and methyl signals but that they were reverse-correlated. A diffusion-ordered spectroscopy (DOSY) analysis showed that the methyl signals are from high molecular weight compounds and that the sugar signals are either from oligosaccharides or ginsenosides. In subgroup 2, composed of Korean Choongcheong ginseng at different ages, we were able to see age-dependent transitions in the score plot. We believe our approach can be applied to detecting the adulteration of ginseng root powders and other herbal products from different origins.