Little information is available to test the various theories which have been proposed to explain the evolution of cauliflory. This study provides such data from observations in Trinidad of the numbers of potential pollinators visiting trunk and canopy flowers and on the size of canopy and trunk flowers and fruits. Subsidiary observations were made on the partitioning of resources between the sexes within flowers. Significantly more potential insect pollinators were trapped around the trunk flowers of two cauliflorous species than were caught around their canopy flowers. Trunk flowers were found to be larger than canopy flowers in four of the seven cauliflorous species studied, but they were smaller in one species. The higher probability of fruit set on trunks than in the canopy may have selected for cauliflory and subsequently increased trunk flower size in insect-pollinated understorey tropical trees. There was a tendency for flowers on the trunk not only to be larger but also to allocate relatively more dry weight to female parts and result in larger fruit than those in the canopy. These observations are consistent with Wallace's theory of the evolution of cauliflory, which argues that the condition arose in the dark understorey of the tropical forest, as a result of selection for trunk flowers which are more apparent to pollinators than are canopy flowers. However, other explanations for the evolution of cauliflory are not precluded as they are not mutually exclusive.