Lipophilic compounds from Korean perilla ( Perilla frutescens ) seeds were characterized to determine the diversity among their phytochemicals and to analyze relationships between their contents. Twenty-four metabolites consisting of policosanol, phytosterol, tocopherol, and fatty acids were identified. The metabolite profiles were subjected to data mining processes, including principal component analysis (PCA), partial least-squares discriminate analysis (PLS-DA), and Pearson's correlation analysis. PLS-DA could distinguish between all cultivars except between Daesil and Daeyeup cultivars. Linolenic acid contents were positively correlated with β-sitosterol (r = 0.8367, P < 0.0001) and γ-tocopherol contents (r = 0. 7201, P < 0.001) among all perilla grains. The Daesil and Daeyeup cultivars appear to be good candidates for future breeding programs because they have simultaneously high linolenic acid, phytosterol, and tocopherol levels. These results demonstrate the use of metabolite profiling as a tool for assessing the quality of food.