Understanding how the scale of pollen transfer determines the outcome of matings is important evolutionarily and a key issue in restoration ecology. We tested the effects of pollen transfer distance for the self-incompatible shrub Grevillea sphacelata using (1) open pollination and transfer among (2) near neighbours, (3) neighbouring subpopulations and (4) populations separated by c. 4 km. We used AFLP markers to test for evidence of genetic differentiation within and among populations. Patterns of seed initiation suggest that open pollinated flowers were pollen limited, although in one subpopulation open seed set was greater than that achieved with pollen from near neighbours or other subpopulations. We detected no other effects of pollen source on seed initiation or seed and seedling development. In contrast, our genetic survey revealed significant spatial autocorrelation to 5 m, moderate differentiation of populations separated by up to 4 km and significant isolation by distance > 16 km. Our data suggest that, although dispersal of pollen may typically be localized, gene flow prevents localized adaptation or co-adaptation and we detected no effects of inbreeding depression. In a restoration context, our results imply that movement of seed between populations separated by 4 km will not have detrimental consequences, despite significant differentiation at neutral genetic markers, and may be beneficial in maintaining genetic diversity and evolutionary potential. © 2013 The Linnean Society of London, Botanical Journal of the Linnean Society, 2013, 173, 290–302.