• Background and Aims: Plants show patterns of spatial genetic differentiation reflecting gene flow mediated by pollen and seed dispersal and genotype × environment interactions. If patterns of genetic structure are determined largely by gene flow then they may be useful in predicting the likelihood of inbreeding or outbreeding depression but should be less useful if there is strong site-specific selection. For many Australian plants little is known about either their population genetics or the effects on mating systems of variation in pollen transfer distances. Experimental pollinations were used to compare the reproductive success of bird-adapted Grevillea mucronulata plants mated with individuals from a range of spatial scales. A hierarchical survey of microsatellite DNA variation was also conducted to describe the scale of population differentiation for neutral markers. • Methods: The effects of four pollen treatments on reproductive performance were compared. These treatments were characterized by transfer of pollen from (a) neighbouring adults; (b) an adjacent cluster of adults (30-50 m distant); (c) a distant cluster (>5 km distant); and (d) open pollination. Sets of 17.9 ± 3·3 leaves from each of 15 clusters of plants were genotyped and spatial autocorrelation and F statistics were used to describe patterns of genetic structure. • Key Results: Grevillea mucronulata displayed evidence of both inbreeding and outbreeding depression, with 'intermediate' pollen producing consistently superior outcomes for most aspects of fitness including seed set, seed size, germination and seedling growth. Significant genotypic structuring was detected within clusters (spatial autocorrelation) and among adjacent clusters and clusters separated by >5 km distance (FST = 0.07 and 0·10). • Conclusions: The superior outcome of intermediate pollen transfer and genetic differentiation of adjacent clusters suggests that G. mucronulata selection disfavours matings among closely and distantly related neighbours. Moreover, the performance of open-pollinated seedlings was poor, implying that current mating patterns are suboptimal. © The Author 2011. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved.