We analyzed the behavior of 3–4-d-old prolarval and 28–33-d-old metalarval American shadAlosa sapidissimain groups of 3–1,000 fish per 22-L glass tank, to determine whether (1) previously described juvenile behavior patterns first develop in larvae, (2) group size or density alters the behavior of larvae, and (3) schooling or other forms of cohesive behavior develop in larvae to promote social interactions. Twelve discrete behaviors or modal action patterns (MAPS) oflarvae were observed at all group sizes; halfthese patterns are unique to larval stages. Conversely, larvae do not develop five previously described juvenile MAPS. Stereotyped metalarval feeding sequences were absent or poorly developed in prolarvae. Group size was directly related to duration of free swimming in water column (metalarvae only) and to frequencies of “proximity to another fish” (all larvae), “contact another fish” (all larvae), and “escape or flee” (all larvae). Age or larval stage significantly affected all swimming-related activities and three feeding behaviors. Larvae foraged and fed independently of one another and used MAPs typical of other larval fishes (“fixate,” “sigmoid,” “lunge,” and “capture”). With one exception (a direct relationship between frequency of food capture and metalarval size), group size and individual size did not significantly affect larval feeding success. Neither schooling nor forms of behavior leading to coordinated group swimming were observed at either larval stage. Larval behavior differed from juvenile behavior in a way that suggests survival in riverine habitats is promoted by behavior that disperses larvae and enables them to function nonsocially.