Connectivity is widely recognised as an important component in developing effective management and conservation strategies. While managers are generally most interested in demographic, rather than genetic connectivity, new analytic approaches are able to provide estimates of both demographic and genetic connectivity measures from genetic data. Combining such genetic data with mathematical models represents a powerful approach for accurately determining patterns of population connectivity. Here, we use microsatellite markers to investigate the genetic population structure of the New Zealand Rock Lobster, Jasus edwardsii, which has one of the longest known larval durations of all marine species (> 2 years), a very large geographic range (> 5500 km), and has been the subject of extensive dispersal modelling. Despite earlier mitochondrial DNA studies finding homogeneous genetic structure, the mathematical model suggests that there are source-sink dynamics for this species. We found evidence of genetic structure in J. edwardsii populations with three distinct genetic groups across New Zealand and a further Australian group; these groups and patterns of gene flow were generally congruent with the earlier mathematical model. Of particular interest was the consistent identification of a self-recruiting population/region from both modelling and genetic approaches. While there is the potential for selection and harvesting to influence the patterns we observed, we believe oceanographic processes are most likely responsible for the genetic structure observed in J. edwardsii. Our results, using a species at the extreme end of the dispersal spectrum demonstrate that source-sink population dynamics may still exist for such species.