Oxygen consumption, CO2 excretion, and nitrogenous waste excretion (75% ammonia-N and 25% urea-N) were measured daily in 4-g rainbow trout over a 15-day starvation period. Oxygen consumption and CO2 excretion declined while N excretion increased transiently in the mid-part of the starvation period but was unchanged from control levels at the end. Component losses (as percentage of total fuel used) of protein, lipid, and carbohydrate were 66.5, 31.1, and 2.4% respectively, as measured from changes in body weight and body composition, the latter relative to a control group at day 0. Instantaneous fuel use, as calculated from the respiratory quotients and nitrogen quotients, indicated that relative protein use rose during starvation, but contributed at most 24% of the aerobic fuel (as carbon). Lipid metabolism fell from about 68 to 37%, and was largely replaced by carbohydrate metabolism which rose from 20 to 37%. We conclude that the two approaches measure different processes, and that the instantaneous method is preferred for physiological studies. The compositional method is influenced by greater error, and measures the fuels depleted, not necessarily burned, because of possible interconversion and excretion of fuels.