Pollen dispersal shapes the local genetic structure of plant populations and determines the opportunity for local selection and genetic drift, but has been well studied in few animal-pollinated plants in tropical rainforests. Here, we characterise pollen movement for an insect-pollinated Neotropical canopy palm, Oenocarpus bataua, and relate these data to adult mating system and population genetic structure. The study covers a 130-ha parcel in which all adult trees (n ¼ 185) were mapped and genotyped at 12 microsatellite loci, allowing us to positively identify the source tree for 90% of pollination events (n ¼ 287 of 318 events). Mating system analysis showed O. bataua was effectively outcrossed (t m ¼ 1.02) with little biparental inbreeding (t m Àt s ¼ À0.005) and an average of 5.4 effective pollen donors (N ep ) per female. Dispersal distances were relatively large for an insect-pollinated species (mean ¼ 303 m, max ¼ 1263 m), and far exceeded nearest-neighbour distances. Dispersal kernel modelling indicated a thin-tailed Weibull distribution offered the best fit to the genetic data, which contrasts with the fat-tailed kernels typically reported for pollen dispersal in trees. Preliminary analyses suggest that our findings may be explained, at least in part, by a relatively diffuse spatial and temporal distribution of flowering trees. Comparison with previously reported estimates of seed movement for O. bataua suggests that pollen and seed dispersal distances may be similar. These findings add to the growing body of information on dispersal in insect-pollinated trees, but underscore the need for continued research on tropical systems in general, and palms in particular. Keywords: Arecaceae; Chocó rainforest; chapil palm; neighbourhood model; phenology; spatial genetic structure; weibull distribution INTRODUCTION Pollination biology is a key determinant of short-term microevolutionary processes in plants. Pollen movement, along with seed dispersal, drives the distribution of genetic diversity in plant populations and is important for generating within-and between-population genetic structure. Wright (1943) recognised that restricted dispersal could lead to population sub-division as an outcome of restricted propagule movement, and introduced the concept of neighbourhood size (N e ) to describe the effective number of randomly mating individuals in a population area. Neighbourhood size has become a common metric of the scale and effectiveness of dispersal, and determining how pollination biology affects N e is a fundamental goal in plant ecology and genetics (Crawford, 1984).The average and maximum distances pollen travels between the pollen source and maternal tree are often used as an index of neighbourhood size by plant biologists (Levin and Kerster, 1971). The focus on distance per se in pollination biology can be related to the early observation that pollen dispersal kernels (that is, the probability density function of propagule dispersal distances from individual plants) typically follow a leptokurtic dis...