Genetic markers have proven useful for determining which sea turtle rookeries contribute to a particular feeding ground. This information is especially relevant when management concerns include anthropogenic mortality of feeding cohorts, and the suspected presence of endangered populations. One such feeding habitat is the Pamlico–Albemarle Estuarine Complex in North Carolina, which was established as an index area in 1995 to monitor population-specific recovery of sea turtles. Pound nets in the study area were surveyed at random from September–December (1995–1997) to enumerate incidental captures of sea turtles as an index of sea turtle abundance. In this study, we estimated the rookery origins of this feeding cohort using both maximum likelihood and Bayesian based stock analysis programs and compare and contrast these different methodologies. The Bayesian methods appear to yield more realistic estimates of percent contribution to the feeding cohort when information regarding relative population sizes was used. Subsequently, we tested for temporal variation in the frequency of mitochondrial DNA haplotypes and resulting estimates of contribution over a 3-year time span. Mixed stock analysis of the combined data indicated that 80 of the individuals originated from the south Florida nesting population, 12 were from the northeast Florida to North Carolina nesting population, 6 from Yucatan, Mexico, and 2 from other rookeries. Although statistically significant shifts in haplotype frequencies were not observed between the three annual sampling periods, estimates of composition indicated subtle differences in the contributions to this foraging area over the sampling period.