The bluegill sunfish, Lepomis macrochirus, is known to select prey on the basis of size. We present evidence that this size selection is related to the optimal allocation of time spent searching for, and handling prey. A model relating search and handling time to energy return is constructed to determine the optimal breadth of diet. Prey are permitted to differ in size and relative abundance. All elements of the model are estimated from experiments with the bluegill feeding on populations constructed from size classes of Daphnia magna. Relative visibility of the different prey sizes markedly affects relative encounter rates or "effective" proportions. Effective proportions are determined empirically from feeding experiments and theoretically from reaction distance in order to correct for this bias. Search time is then manipulated by varying absolute abundance of prey. At low absolute abundance, prey of different size are eaten as encountered. As prey abundance is increased, size classes are dropped sequentially from the diet in accordance with the theory. Search and handling times are estimated from these experiments and quantitative comparisons with the model indicate these changes in diet maximize return with respect to time spent foraging.