The sika deer population on Yakushima Island exhibits high genetic diversity despite the small size of the island. We hypothesized that the high genetic diversity of the population had been maintained by the population structure, which included several subpopulations among which gene flow was limited. To test this hypothesis, we analyzed the spatial genetic structure of the population using 12 microsatellite loci. Two and four subpopulations were detected by STRUCTURE (stN and stS) and GENELAND (glN, glE, glS, and glW) software, respectively. The basic genetic structure assigned by STRUCTURE was supported by GENELAND, while stN and stS were further separated into two subpopulations (glN and glE; glS and glW, respectively) by GENELAND. All pairwise genetic differentiations between the two and four subpopulations were significant. These results demonstrated that the Yakushima population was structured into genetically distinct subpopulations. Although the location of the western boundary between stN and stS corresponded with a large river, no landscape or biological feature could be identified for the eastern boundary. We discussed the relationships between the genetic structure and management units designed by the local government and concluded that the Yakushima population should be managed based on their spatial population structures considering multiple time scales.