Habitat edges are considered to have an important role in determining the abundance of deer in forest landscapes, but to our knowledge there are few lines of evidence indicating that forest edge enhances the vital rate of deer. We examined pregnancy of female sika deer in Boso peninsula, central Japan, and explored how forest edges, food availability in forests, and local population density influence the pregnancy rate of sika deer. Local deer density was estimated by the number of fecal pellets, and food availability in forests was estimated by combining GIS data of vegetation distribution and the relationship between vegetation biomass and local deer density. Forest edge length was also determined by GIS data. Model selection was performed with multiple logistic regression analyses using the AIC to find the best model for accounting for the observed variation in pregnancy rates of the deer. Multiple logistic regression analysis showed that the length of forest edge had a positive effect on the pregnancy rate of females, whereas food availability in forests and local deer density had little effect. This forest edge effect was detected in a 100-200-m radius from deer captured locations, indicating that deer pregnancy is primarily determined by habitat quality within a 10-ha area. This result was confirmed by tracking females with GPS telemetry, which found that the core areas of the home range were less than 12 ha. The positive effect of edges and the lack of density dependence could be a result of high plant productivity in open environments that produces forages not depleted by high deer densities. Our results support the view that land management is the cause of the current problem of deer overabundance.