Analyses of stomach contents showed that the kinds of prey eaten by brook trout (Salvelinus fontinalis), cutthroat trout (Salmo clarki), and rainbow trout (Salmo gairdneri) were seldom distributed at random among the individuals. Repeated observation of food eaten by individuals in a stream and ponds showed that prey types were eaten in proportions which were characteristic for an individual.Specialization occurred on several different kinds of prey. Although the degree of specialization was higher during shorter intervals, the data suggested that some specialization persisted for half a year. There were no striking correlations between degree of specialization and other individual properties such as size, growth rate, weight of food, number of food items, previous specialization, or area of recapture.In addition to the observations on trout in relatively undisturbed habitats, a field experiment was conducted using laboratory-reared rainbow trout held in small ponds. The food of each trout in the experiment was sampled repeatedly. In analysis of variance, interaction among the individuals and kinds of prey eaten showed that food specialization occurred. Both the absolute and relative abundance of potential prey were constant during the experiment.