Pollination success in female trees was determined for a population of Neolitsea dealbata (R. Br.) Merr., a locally abundant dioecious tree pollinated by small, unspecialized insects in northern Queensland rain forest, Australia. The population consisted of a clustered group of trees with a mean male-to-female distance of 4.5 m and more isolated individuals, including females more than 90 m away from the nearest pollen source. A map of all reproductive trees was produced to determine accurate male-to-female distances. The size of the pollen source available to females was defined as a function of the distance to the nearest ten male trees and their sizes (male neighbourhood index). The rate of pollen movement to females was measured by counting pollen tubes (and the number of tubes per style) in female trees 6 days after the commencement of population flowering. The pollination rate decreased steeply to less than half when the nearest male was only 6.5 m away. Although pollen reached a female 330 m away from the nearest pollen source, only 10% of receptive flowers had been pollinated. The short flowering period (2-3 weeks) combined with the the slow rate of pollen movement means that a large proportion of flowers in isolated trees are unpollinated, confirming an earlier finding that isolated females set fewer fruits than gregarious females. The reliability of pollen transfer to females was determined by quantifying insects and their pollen loads trapped at female trees with a range of male neighbourhood indices. Quantities of insects and pollen were significantly correlated with the size of the male neighbourhood index, indicating a strong density-dependent response by vectors to flowering. Pollen was also collected from insect visitors to non-flowering trees. Females with large male neighbourhood indices received more pollen than non-flowering trees with equivalent male neighbourhood indices. However, when the male neighbourhood indices were small for both female and non-flowering trees, the changces of pollinators encountering female and non-flowering trees were similar, suggesting random movements of pollinators in sparse-flowering sub-populations. The dioecious breeding system, brief, synchronous flowering period, clustered population structure and random, opportunistic foraging behaviour of vectors interacted in a way that reduced reproduction in relatively isolated trees. These results demonstrate a mechanism for differential breeding success between trees in natural populations and emphasize the possible impact of logging regimes on pollen flow between trees. Large interconspecific distances in species-rich environments may have been a factor in the selection for synchronous flowering between trees in outcrossing tree species with generalist insect pollinators.