A mechanistic model is developed to determine the habitat needs of drift-feeding stream salmonids from the direct cause-and- effect relationships of environmental and physiological variables on net energy intake (NEI). The model determines NEI by subtracting energy costs (basal metabolism , swimming cost, digestion cost) and losses (egestion and excretion) from the gross energy intake obtained as a result of simulated prey capture. The prey capture portion of the model utilizes components o f the predation model of C.S. Holling and the prey capture model of N.F. Hughes and L.M. Dill to determine the rate of prey capture (gross energy intake) as a function of fish size, water velocity, water depth, water temperature, and the amount of drift. Physiological input parameters for the model are estimated from the literature. Two separate validation tests of the model's ability to predict stream habitat use of trout, primarily cutthroat trout (Oncorhynchus clarki), in St. Charles Creek, Idaho, are presented. In both cases, the NEI model closely predicts the stream habitat that different size classes of fish utilize. The validation tests provide strong evidence that drift-feeding fish utilize stream habitats that provide high rates of NEI as determined by the model. Sensitivity and simulation analyses of the model are used to identify the most important input parameters and to illustrate in terms of energetics why drift-feeding fish utilize various habitats. Model simulations explain why fish utilize deeper and faster habitats as they get larger and why they utilize slower habitats in the winter. In addition, it is shown that streams with high drift rates should theoretically provide more usable salmonid habitat than similar streams with lower drift rates.