Predation on American shad (Alosa sapidissima) larvae within the first two hours of release was examined from 1989 to 1992 on 31 occasions at stocking sites in the Susquehanna River basin. Twenty-two fish species consumed shad larvae; the dominant predators were spotfin shiner (Cyprinella spiloptera), mimic shiner (Notropis volucellus) and juvenile smallmouth bass (Micropterus dolomieu). The number of shad larvae found in predator stomachs ranged from 0 to 900. Mortality of shad larvae at the stocking site was usually less than 2%. The greatest mortality (9.6%) occurred at the highest stocking level (1.5 million larvae). Highly variable predation rates and release levels of shad insufficient to achieve predator satiation hindered the ability to determine a specific type of functional response of predators. Predator numbers increased with stocking density, indicating short-term aggregation at the release site. Because of practical problems associated with releasing the large numbers of larvae that would be required to satiate predators, routine stocking at these levels is probably unreasonable. Releases of 400,000 to 700,000 larvae may reduce predation by offsetting depensatory mechanisms that operate on small releases and the effects of increased predation due to predator aggregation on large releases. Night stocking may reduce predation on larval shad at the release site.