Because pollinator behaviour within plants affects the mating outcomes of animal-pollinated plants, plants may regulate the pollinator behaviour by controlling floral design and display to maximize pollination efficiency. We assessed how extent of the display size, inflorescence architecture, and nectar distribution within inflorescences alter bumblebee behaviour (number of flowers visited, time of stay, and nectar intake efficiency) under controlled conditions using artificial inflorescences. The effect of display size on bumblebee attraction was also measured. Eighteen types of artificial inflorescences, two display sizes (large and small) × three architectures (panicle, umbel, and raceme) × three nectar distributions (constant nectar content in every flower, gradual changes in nectar content, and random distribution of empty and rewarded flowers) were visited by bumblebees. The mean nectar amount per flower was set as 3 μL in 2003 and 1 μL in 2004. Although the inflorescence architecture had little effect on bumblebee behaviour, the random nectar distribution decreased the number of successive flower visits (both 3 μL and 1 μL) and staying time (only 3 μL) during a single visit and increased the foraging efficiency (both 3 μL and 1 μL). These features may contribute to decreases in the geitonogamous pollination and pollen loss during pollinator movements. Display size did not influence the attraction of bumblebees in this study. Our results suggest that nectar distribution within inflorescences can be a strategy for efficient pollination by manipulating pollinator behaviour.