The raccoon (Procyon lotor) is an ecologically important mesopredator that threatens at least 11 species of beach-nesting and colonial waterbirds on the Virginia barrier islands. An understanding of population dynamics, reproductive sources and sinks, and dispersal pathways among the islands and the adjacent Delmarva Peninsula mainland will help prioritize allocation of scarce resources for managing these raccoon populations. We characterized these metapopulation processes by examining variation across mitochondrial (mtDNA) and nuclear DNA (microsatellite) markers. We sequenced a 515-base pair fragment of mtDNA (containing the 5′-end of the cytochrome-b gene, the 3′-end of the D-loop, and the intervening tRNA genes) in 164 animals from 22 localities. We detected 7 unique mtDNA sequences, distributed along a presumably long-term, north–south genetic gradient on the islands and on the mainland, reflecting the temporal sequence of colonization of these areas. We also genotyped 13 nuclear microsatellite loci in 314 individuals from 24 localities, and we found complex patterns of spatial population structure, dispersal, and gene flow among island and mainland localities. Whereas mainland localities showed considerable admixture, suggesting multiple waves of colonization, most islands showed little admixture, suggesting single founding events and relative isolation since founding. Islands formed 2 genetically cohesive groups, 1 in the north and another in the south. Based on patterns of genetic variation, estimates of effective population size, and immigration–emigration rates, Parramore and Revel islands serve as the major demographic source of raccoons on the northern islands, whereas Smith Island is the major source of raccoons on the southern islands. Most inferred movements of raccoons occurred among adjacent groups of islands that are interconnected by marsh and relatively shallow, narrow, open-water channels. These genetic results are consistent with our empirical studies of raccoon movement and support predictions from a cost-distance model of raccoon movement, which assumes that gene flow is primarily constrained by a combination of spatial distance and landscape resistance. Our landscape genetic analyses indicated that: 1) raccoons occur in a series of semi-independent local populations that exhibit significant genetic structure and are characterized by historical and ongoing colonization events, 2) island populations exhibit evidence of source–sink dynamics in patterns of variation in genetic diversity, effective population size, and emigration–immigration rates, 3) the Delmarva Peninsula mainland is not a major source of raccoons dispersing to most of the islands, and 4) individuals dispersed mostly among nearby islands and along least-cost pathways of landscape resistance, in agreement with predictions based on cost-distance models of raccoon movement in this system.