In mass culture of Pacific bluefin tuna Thunnus orientalis, a marked growth variation is observed after they start feeding at 6–7 mm in body length (BL) on yolk-sac larvae of other species, and the growth variation in tuna larvae is a factor leading to the prevalence of cannibalism. To examine the relationship between prey utilization and growth variation, nitrogen stable isotope ratios (δ15N) of individual larvae were analysed. A prey switch experiment was conducted under two different feeding regimes: a group fed rotifers (rotifer fed group), and a group fed yolk-sac larvae of spangled emperor, Lethrinus nebulosus (fish fed group) from 15 days after hatching (6.87 mm BL). The fish fed group showed significantly higher growth than the rotifer fed group. Changes in the δ15N of the fish fed group were expressed as an exponential model and showed different patterns from those of the rotifer fed group. The δ15N of fast-growing tuna larvae collected in an actual mass culture tank after the feeding of yolk-sac larvae was significantly higher than those of the slow-growing larvae, indicating that slow glowing larvae depended largely on rotifers rather than the yolk-sac larvae.