E xogenous protein in the absence of other calories can cause protein-sparing, but the mechanisms involved are controversial. It has been postulated that low insulin and high fat-derived substrate levels are necessary and sufficient conditions for such protein-sparing. We therefore established such conditions with differing protocols of protein input to define the role of protein input in mediating the response. Three groups of obese, nondiabetic subjects received the following diets: (1) 82.5±1.0 g protein/day (400 cal/day) for 21 days, n = 7; (2) the same, but as a refeeding diet for 7 days after 21-28 days of total fasts, n = 7; and (3) commencing with the same input, but with daily stepwise decrements over 14 days to 19.4±2.2 g/day, then maintained an additional 7 days, n = 4. Diet 3 gave approximately the amount and pattern of protein lost during total fasting.The circulating hormone and substrate responses of diets 1 and 3 were comparable and resembled those of total fasts, in that plasma glucose and insulin fell and free fatty acids rose. Blood levels of alanine, pyruvate, and other glucogenic amino acids fell and blood levels of branched-chain amino acids rose transiently. Blood 3-hydroxybutyrate levels and urinary excretion were greater in diet 3 than diet 1, but less than in total fasting. Nitrogen balance in diet 1 was transiently negative, but in equilibrium from 12 to 21 days. In diet 3, it was constantly negative at −6 g/day, the values also observed at 21 days of fasting. Mean 3-methylhistidine excretion decreased by 170 μmol/day in diet 1 and 107 μmol/day in diet 3, reflecting decreased muscle protein catabolism. The refed, protein-depleted subjects, diet 2, showed an increase in plasma glucose without alteration in insulin levels. Free fatty acid and ketone body levels decreased to those of the steady state observed in diet 1. Glucogenic and branched-chain amino acids decreased transiently. Nitrogen balance became positive, and the low 3-methylhistidine excretion increased by 152 μmol/day.The differing responses of nitrogen balance could not be accounted for on the basis of levels of insulin or of fat-derived substrates. The primary determinants of the protein-sparing observed appeared to be the protein supply itself, and the magnitude of the decrease in endogenous protein catabolism. The positive balance on refeeding after prior depletion of protein stores was likely due to the exogenous supply, combined with decreased catabolism and considerably increased reutilization.