Simple Summary: Herbivorous mammals are considered to eat plants with high nutrition but to avoid those with harmful chemical components. Leaf chemical components differ between plant species, conspecific individuals, leaves of the same individual, and the parts of a single leaf; therefore, food selection by herbivores (particularly browsers) occurs in these hierarchical structures. However, the effects of plant chemicals on food selectivity are not well known in the tree-leaf-eating mammals such as arboreal primates, rodents, and marsupials, as compared with other mammals eating herbaceous plants, seeds, nuts, and fruits. Moreover, the effects of the microscale chemical distribution within the single leaf on their partial feeding have been little examined. The Japanese giant flying squirrel (Petaurista leucogenys) is an arboreal herbivorous mammal that feeds on mostly tree leaves. In our study site, the three tree species, evergreen Phonitia serratifolia, deciduous Quercus acutissima, and evergreen Q. sessilifolia, were used by this flying squirrel as main food. In comparisons of feeding preference and chemical (phenolic, glucose, and water) contents of leaves among these tree species, leaf glucose concentration was an important factor determining which species of tree leaves they eat. In comparisons between favorite parts of consumption and chemical contents within the single leaf, the central consumption was frequent when the water contents were higher at the leaf center than the margin. Phenolic contents in the central and marginal parts of the single leaf differed among the three tree species, but clear tendency to avoid eating the part with high phenolic contents was not detected. To examine the effect of leaf chemical composition on selective herbivory by the Japanese giant flying squirrels (Petaurista leucogenys), we measured and compared the total phenolic, glucose, and water contents of leaves among their main food tree species, deciduous Quercus acutissima, and evergreen Q. sessilifolia and Phonitia serratifolia. Leaves of these three tree species were available in the warm season (April to October), but the flying squirrels mostly preferred the leaves of Q. acutissima, having higher glucose and water contents than those of the other two tree species. In the cold season (November to the next March), the two evergreen tree species were available, and the flying squirrels used both leaves without any apparent influence of the chemical compositions. On the other hand, the favorite parts of a single leaf differed among the three tree species. Flying squirrels dropped the individual leaves after partial consumption. Their feeding marks on the dropped leaves were distinguished into four types: apical, basal, central, and marginal parts of consumption. The basal parts of consumption were most frequent in Q. acutissima leaves in which more water was contained at the basal part, and the central part consumption followed, which may be related to a lower phenolic content and more glucose and water at the leaf center than its margin. In contrast, the apically consumed leaves dominated in Q. sessilifolia, with relatively homogeneous leaf chemical distribution except for more water at the center. In P. serratifolia, leaves consumed at the center were frequent, but those with marginal consumption were also observed, which may be related to its specific chemical distribution with less phenolics and more glucose at the leaf margin. Thus, the chemical distributions within the single leaf differ among tree species, and the flying squirrel's selectivity of the tree species and the part of each leaf depends partly on the relative compositions of preferable glucose and water and unpreferable phenolics. [ABSTRACT FROM AUTHOR]