Despite intense interest in conservation of marine turtles, spatial ecology during the oceanic juvenile phase remains relatively unknown. Here, we used mixed stock analysis and examination of oceanic drift to elucidate movements of hawksbill turtles (Eretmochelys imbricata) and address management implications within the Caribbean. Among samples collected from 92 neritic juvenile hawksbills in the Cayman Islands we detected 11 mtDNA control region haplotypes. To estimate contributions to the aggregation, we performed 'many-to-many' mixed stock analysis, incorporating published hawksbill genetic and population data. The Cayman Islands aggregation represents a diverse mixed stock: potentially contributing source rookeries spanned the Caribbean basin, delineating a scale of recruitment of 200-2500 km. As hawksbills undergo an extended phase of oceanic dispersal, ocean currents may drive patterns of genetic diversity observed on foraging aggregations. Therefore, using high-resolution Aviso ocean current data, we modelled movement of particles representing passively drifting oceanic juvenile hawksbills. Putative distribution patterns varied markedly by origin: particles from many rookeries were broadly distributed across the region, while others would appear to become entrained in local gyres. Overall, we detected a significant correlation between genetic profiles of foraging aggregations and patterns of particle distribution produced by a hatchling drift model (Mantel test, r = 0.77, P < 0.001; linear regression, r = 0.83, P < 0.001). Our results indicate that although there is a high degree of mixing across the Caribbean (a 'turtle soup'), current patterns play a substantial role in determining genetic structure of foraging aggregations (forming turtle groups). Thus, for marine turtles and other widely distributed marine species, integration of genetic and oceanographic data may enhance understanding of population connectivity and management requirements.