Natal dispersal can have important effects on mammal population structure and dynamics following a local population crash. Such dispersal is of practical importance when applied to the control of pest species because dispersal may significantly, and undesirably, reduce the population recovery time following a control operation. The relative dispersal rate of the sexes is also critical because that too will affect the rate of population increase. Here, we describe a field experiment in which we reduce the density of two populations of the Australian brushtail possum, and use genetic similarity, as estimated by minisatellite DNA profiles, to investigate dispersal in the original (undisturbed) and recovering populations. Our results show that the genetic similarity within the undisturbed populations was lower between males than between females. Conversely, the genetic similarities between males and females in the two recovering populations were not significantly different, while relatedness among males was significantly higher in the recovering populations when compared with those in the pre-removal populations. These data indicate two important characteristics of dispersal in possums: (i) that dispersal in established populations is sex biased towards males; and (ii) that within the first 3 years following population control, 'the vacuum effect', whereby individuals from areas adjacent to a control area expand their home range and invade the depopulated area, is the most important factor in the re-colonization process for possums. We found no evidence that the mating system, which is polygynous, varied when the density was markedly reduced. These results indicate that drastic reductions in population density by conventional control will not affect the rate of spread of biological control agents that rely on sexual transmission for dissemination.