The proximate and ultimate causes of dispersal in semelparous carnivorous marsupials (Phascogalinae) have previously been hypothesized to be maternal aggression and inbreeding avoidance, respectively. This study tests these hypotheses by exposing 26 litters of Phascogale tapoatafa to a diverse range of social and environmental conditions that potentially affect dispersal (e.g. supplemental feeding, post-weaning desertion by the mother, orphaning, and release of subadults into unoccupied habitat). The mean dispersal age was 162 ± 5.6 d, which is about 3 wk after weaning is complete. Juvenile dispersal was strongly male biased under all conditions, suggesting that extrinsic proximate causes do not adequately account for male emigration. Home range establishment by males was contingent on the presence of females. Half of the monitored daughters were philopatric, and others typically settled adjacent to the natal site, thus possibly enhancing their reproductive potential by occupying an area of known resource quality. Because philopatry increases the risk of incest, females may be selected to preferentially mate with unrelated males (immigrants), when they are available, to avoid inbreeding. If so, the presence of immigrant males would reduce the probability of locally born, related males reproducing at their natal site. Thus inbreeding avoidance by females may create local mate competition among males and select for male dispersal. Emigration also ensured that males avoided inbreeding, but, if they dispersed into unoccupied habitat, male P. tapoatafa often returned to the natal area. This 'boomerang strategy' of returning to mate with related females suggests that, in the absence of conspecifics along the dispersal path of a male, mate competition will be weak at the natal site and female mate choice will not preclude related males. Thus while inbreeding avoidance by either or both sexes is perhaps the most parsimonious explanation of male-biased emigration, dispersal patterns were apparently strongly influenced by additional factors, so that the ultimate causation of the dispersal regime may be more complex.