In the present study, non-aqueous fractionation (NAQF) and GC-MS were used to obtain a spatially resolved view of metabolism in mature leaves of soybean (Glycinemax Merr.). NAQF of lyophilized soybean leaves was performed using CCl4-n-heptane and ultracentrifugation that yielded a gradient comprised of six fractions. Chlorophyll content, and marker enzyme activities, phosphoenolpyruvate carboxylase (PEPC) and α-mannosidase, were utilized as stroma, cytosol and vacuole markers, respectively. GC-MS analyses of each fraction resulted in the identification of around 100 different metabolites. The distribution of these identified compounds showed a decreasing order from the vacuole to cytosol to chloroplast stroma. In other words, a greater number of identified compounds were found in the vacuole when compared to the cytosol or stroma. Levels of sugars, organic acids and fatty acids showed greater relative abundances in the vacuole with 50, 55, and 50% of the respective pools. A greater relative abundance of amino acids was observed in the cytosol where 45% of the total of amino acids content was recorded. The relatively large pool of sugars and phenolic acids in the vacuole compartment implies high levels of starch metabolism and phenylpropanoid biosynthesis. The low amino acids pool, on the other hand, suggests low nitrogen accumulation in the leaves of soybean. Hierarchical cluster analysis on the most abundant metabolites revealed three clusters containing 10, 20, and 2 of the 32 selected metabolites. The data were discussed in term of NAQF and GC-MS analysis of soybean mature leaves, and also in term of distribution and compartmentation of metabolites at subcellular levels.