Metabolite fingerprinting was applied in the attempt to evaluate the quality of Angelica acutiloba dried roots (Yamato-toki). A pyrolyser coupled to a gas chromatography mass spectrometer (PY-GC-MS) was used to obtain higher chemical universality by analyzing whole compounds including high molecular weight metabolites. The machine was relatively fast and easy to use, with no sample preparation procedure required. Multivariate pattern recognition methods, specifically principal component analysis (PCA) and partial least square discriminant analysis (PLS-DA), were successful in discriminating various toki samples. In addition, an enhanced understanding of the dominant relationship of cultivation area to quality evaluation was conceptualized and therefore applied to the construction of a PLS-DA classification model which provided the basis for accurate and reliable predictivity. (C) 2009, The Society for Biotechnology, Japan. All rights reserved.