An experiment was conducted in which the metabolic utilization of energy was measured in individually penned pigs from seven groups that differed in genotype and(or) sex and ranged in body weight between 20 and 107 kg. The animals were fed a diet containing, on a DM basis, 14.7 MJ ME and at least 21% CP. Heat production was measured in an open-circuit calorimeter, and energy, nitrogen, and fat balances were determined at regular intervals over the growing period; a total of 177 measurements were performed. Body composition of the animals was measured by serial slaughter, and these data were used for estimating the body composition of an animal at a given weight through allometric regression. A factorial analysis procedure was used to estimate the utilization of ME by regressing the ME intake on the observed protein and lipid deposition rates. The intercept of this equation is the maintenance energy requirement (MEm) and was represented either as a function of body weight with group-specific parameters (MEm = a(i) BWb) or as a function of the muscle and visceral mass with an additional additive group effect (MEm = aM muscle(b) + a(v) viscera(b) + G(i)). With BW as dependent variable, the exponent b was close to .60 and differed significantly from .75. The regression coefficient (a(i)) averaged 1.02 MJ ME/kg.60 but it was different for most groups, indicating that different groups of animals have different maintenance requirements. Fixing the exponent to .75 consistently underestimated the maintenance requirement. When the exponent b was not fixed to .75 but estimated, the partial efficiencies for protein and lipid deposition were .62 and .84, respectively. Body muscle and visceral mass could explain a large part of the variation in MEm. Viscera contributed three times more to MEm (per kilogram of mass raised to the .70 power) than did muscle. Even though the muscle mass exceeds to a large extent the visceral mass in animals, the contribution of muscle to MEm was lower than that of viscera for most groups.