A general understanding of the factors that regulate populations remains a central goal for ecologists, and species that increase rapidly to plague numbers followed by equally sudden crashes in abundance are particularly intriguing. Throughout central Australia, the long-haired rat, Rattus villosissimus, is largely absent in dry times, but can irrupt into plague numbers after heavy rains. Using historical data we first relate the occurrence of plagues to exceptionally high rainfalls, and using live-trapping records from rat irruptions in 1991 and 2011 in the eastern Simpson Desert, we then investigate the population structure at the plague front. In doing so, we ask if long-haired rat plagues irrupt from multiple refugia, if young males are first to disperse and whether the rate of dispersal is rapid. Annual rainfall in the year preceding a plague successfully predicted the probability of an outbreak, with an 80% chance of a plague occurring after an annual rainfall of 750 mm. Trapping grids closest to drainage lines to the south of the study area had greater captures of long-haired rats. Enhanced rat numbers in proximity to drainage areas suggested that the irruptive process unfolded from the south and demonstrated the existence of multiple refugia. In contrast to the resident fitness hypothesis, there was no difference in captures between the sexes or age-classes of long-haired rats. Heavier animals were caught at the invasion front, presumably because they could travel larger distances than smaller individuals, suggesting that larger individuals were maximizing their fitness by dispersing first to new resources. Captures of long-haired rats were highest at the beginning of the irruption in both 1991 and 2011, suggesting rapid colonization. Although rare in occurrence, rodent population irruptions and the resulting dispersal across large areas, which may persist for a few years, can thus extend the domain of influence of exceptional rain years well into the future.