Congo Cay, U.S. Virgin Islands, has high value for breeding seabirds and is a potential reintroduction site for the endangered Virgin Islands tree boa (Epicrates monensis granti). However, introduced ship rats (Rattus rattus) undermine its conservation value. Three unsuccessful eradication attempts have been conducted since 1990, with the latest in 2006; rats were trapped 1.5 years later. We examined microsatellite DNA and mitochondrial DNA sequences of ship rats from Congo Cay and three other nearby cays to determine if rats found after the most recent eradication effort were surviving individuals or reinvaders from neighboring cays; we had no pre-eradication samples. Only one mitochondrial haplotype was present, implying that historically there was a single invasion or if multiple invasions, rats came from a single source with limited haplotype diversity. Low genetic variation on Congo Cay suggested either a population bottleneck resulting from survivors or a founder event resulting from invaders. F ST estimates, cluster distances, migrant detections, and factorial correspondence analysis indicated low but meaningful levels of gene flow between Congo and Lovango Cays and between Mingo and Grass Cays. Except for two alleles, all other alleles found on Congo were also present on Lovango. Without pre-eradication samples we could not eliminate the possibility of survivors from a failed eradication. However, our data suggest reinvasion from Lovango Cay was likely and that future eradication efforts should consider both pairs of cays as eradication units. Cay juxtaposition and orientation along with ocean currents may explain rat movement, or lack thereof, among these cays.