ABSTRACT: The Pacific oyster Crassostrea gigas was, for decades, massively introduced to North America from Japan and established large, self-recruiting populations in the Pacific Northwest of the USA and Canada. A previous study of mtDNA variation revealed little population genetic structure among populations from British Columbia and Washington State. Here, we used samples from that study, more recent samples from 2 of the same localities, and 2 additional samples, including 1 from Japan, to investigate spatial and temporal genetic variation at 52 mapped, coding, single-nucleotide polymorphisms (SNPs) assayed by high-resolution melting (HRM). Little variation was detected among North American populations, which, as a group, are distinct, perhaps adaptively so, from oysters in Hiroshima, Japan. However, significant excesses of heterozygotes with respect to random mating expectations and of pairwise linkage disequilibria revealed that North American populations are not in Hardy-Weinberg (random mating) equilibrium. Moreover, genetic changes over 10 to 21 yr in 2 localities are substantial, despite high gene flow, and are as large as spatial variance per generation. These results caution against basing connectivity or seascape genetic analyses on snapshots of spatial population structure in high gene-flow species. Because migration and selection are ruled out as causes of temporal genetic change, random genetic drift is the most parsimonious explanation. This implies effective population sizes (N_e) of hundreds to a few thousands, orders of magnitude smaller than the natural abundance (N) of this oyster. These low N_e:N ratios are compatible with the hypothesis of sweepstakes reproductive success.