Oaks are a dominant component of many North American forests, yet in many areas oak seedling production is declining. Oaks are generally thought to be highly dispersal limited, which could hamper reaching scarce recruitment sites and limit oaks’ ability to respond to climate change via migration or local adaptation. In this study, we apply a Bayesian parentage model developed for monoecious plants to two populations of red oak (Q. rubra) in North Carolina: in the Piedmont (12 ha) and the Coweeta LTER in the southern Appalachians (7.5 ha). The model incorporates genetic data, as well as location and fecundity data. Both plots are located in diverse secondary hardwood forests where oaks are abundant in the canopy; however, they differ in a number of other ecological factors. We ask whether the scale of effective seed and pollen movement differs between these populations and, if so, this difference can be explained by site history and plant-animal interactions. Parentage assignment was surprisingly low, with as many as 10% of seedlings lacking an in-plot parent. Among consistent matches, the average parent-offspring distance in the Piedmont was >70 m. Although at Coweeta a lower percentage of adult trees had been genotyped a greater proportion of seedlings were matched to in-plot adults, and the average dispersal distance for consistent matches was <40 m. This difference is most likely due to biotic interactions; disperser activity is higher at the Piedmont site, but so is herbivory which, when density dependent, can lead to increased apparent dispersal distances. Patterns of spatial genetic structure are consistent with the site history. Based on effective dispersal distances, the Piedmont population would yield higher potential migration rates. Results suggest that some oak populations are less dispersal limited than has been thought, suggesting that regeneration and population expansion are best promoted by favorable disturbance frequencies and reduced herbivory.