The field of ecological restoration is growing rapidly, and the sourcing of suitable seed is a major issue. Information on the population genetic structure of a species can provide valuable information to aid in defining seed collection zones. For a practical contribution from genetics, a rapid approach to delineating seed collection zones using genetic markers (amplified fragment length polymorphisms [AFLPs]) has been developed. Here, we test the effects of sampling regime on the efficacy of this method. Genetic data were collected for an outcrossing seeder, Daviesia divaricata ssp. divaricata, an important species in urban bushland restoration in Perth, Western Australia. The effect of sample size and number of AFLP markers on estimates of genetic variation and population structure was examined in relation to implications for sourcing material for restoration. Three different sample sizes were used (n= 8, 15, and 30) from six urban bushland remnants. High levels of genetic diversity were observed in D. divaricata (87.4% polymorphic markers), with significant population differentiation detected among sampled populations (ΘB= 0.1386, p < 0.001). Although sample size does not appear to affect the spatial pattern in principle co-ordinates analysis (PCA) plots, the number of polymorphic loci increased with sample size and estimates of population subdivision (FST and ΘB) and associated confidence intervals decreased with increasing sample size. We recommend using a minimum of 30 plants for sourcing seed for restoration projects.