Hormonally-induced fertility control is increasingly used to manage overabundant wildlife where political or ethical considerations exclude culling or translocations. By altering reproductive rates, fertility control risks accelerating the loss of genetic variation from isolated wildlife populations if control is biased against particular genotypes. This is most likely in spatially structured populations, but information on patterns of genetic exchange can be difficult to ascertain by observation or trapping. Fertility control was proposed to manage an isolated but overabundant population of the black-flanked rock-wallaby, Petrogale lateralis lateralis, at Mt Caroline Nature Reserve, south-western Australia. This species typically inhabits large granite outcrops, where refuges within outcrops are often patchily distributed, and there is strong potential for population structure. We used fine-scale analysis of microsatellite DNA variation to establish the pre-fertility control extent of population structuring and the relative extent of gene flow in males and females in an isolated population of black-flanked rock-wallabies. Bayesian model-based clustering and fixation statistics revealed that overall gene flow is extensive, but two areas of refuge were genetically distinct from distant neighbours. Spatial autocorrelation analyses indicated that female-mediated gene flow was less extensive than male-mediated gene flow. Our results indicate that it would be desirable to apply fertility control to female rock wallabies within defined genetic isolates, but fertility control could be applied in a spatially random manner to males.