Predator-specific growth-selective predation on larval Japanese anchovy Engraulis japonicus was demonstrated by comparing growth rates between the larvae ingested by predators and the larvae from the corresponding original populations through otolith microstructure analysis, based on original data and reanalyzed data from previous studies. Ingested larvae from the stomachs of small pelagic predators (juvenile Japanese anchovy, round herring Etrumeus teres, jack mackerel Trachurus japonicus and white croaker Pennahia argentatus) had significantly lower growth rates than the larvae from the original populations in general. For large piscivorous predators (sea bass Lateolabrax japonicus, greater amberjack Seriola dumerili and skipjack tuna Katsuwonus pelamis), no measurable differences in the growth rates were observed between ingested larvae and larvae from the original populations. Small pelagic fish were therefore identified as growth-selective preda- tors, whereas large piscivorous fish were identified as non-growth-selective predators. Exponential declines in the relative predation mortalities of larvae with higher growth rates suggest the potential for growth rate to exert a great effect on recruitment variability. However, the predator field would regulate selection for growth characteristics of survivors.