The fruit set ratios of three perennial herbs were investigated to study the effect of forest fragmentation on the pollinator-plant relationship. The fruit set ratios of Corydalis ambigua, Polygonatum odoratum var. maximowiczii and Aconitum yesoense were compared among 14 forestislands, 5 of which were situated in residential areas and 9 in cultivated land. These flowers are outcrossers and pollinated by bumblebees. The average fruit set ratio of C. ambigua, which flowers in April, was lower in residential areas than in cultivated land. In contrast, the average fruit set ratio of P. odoratum, which flowers in June, was high both in residential and cultivated areas. In the case of A. yesoense, which flowers in August, the ratio was also high both in the residential and cultivated areas. Artificial pollination raised the fruit set ratio of C. ambigua in the residential area. The pollen removal ratio (number of flowers with pollen removed/number of flowers observed) was investigated as an indicator of the frequency of pollinator visits on C. ambigua, and showed that pollinator visits were scarce in the forest-islands with a low fruit set ratio. The low fruit set ratio of C. ambigua in residential areas was thus caused by pollinator limitation. Few pollinator visits might be caused by a shortage of flowers that supply the food to pollinators in residential areas.