Using indirect calorimetry the energy expenditure of healthy subjects was investigated as complete 24-hour balance measurements. The respective data were evaluated with regard to level, variation and normalizing of heat production (HP) and maintenance requirement. Diets were composed of customary ingredients and the intake of energy maintained constant slightly below maintenance requirement during the 14-day experimental period. Faeces and urine were collected for the final 7 days. Exchange of oxygen and carbon dioxide was measured during the last 48 h while the subjects stayed in an open-circuit respiration chamber, furnished like a small living-room. The daily energy balance averaged -0.38 MJ ( ± 0.73). For data evaluation the HP of each subject was corrected to zero energy balance, assuming a calorigenic effect of 10% for mixed diets. The resulting energy requirement of metabolizable energy for maintenance at energy equilibrium amounted to 136 ( ± 17) MJ/kg body weight and 383 ( ± 45) kJ/(kg body weight)0’75, respectively. Scaling the body weight to the ¾ power diminished slightly the coefficient of variation. The observed total variance of the HP was mainly contributed by the subjects per se (94%), whereas only a small amount was related to day-to-day variability (6%; measurement noise and intraindividual variability). In terms of coefficient of variation the respective values were 11.5 and 2.8%. Comparable assessments with sows showed nearly the same results on animal variance and day-to-day variability. The daily time pattern of HP showed two maxima and two minima. In the morning and evening the HP exceeded the average 24 h value by 19 and 14%, respectively. During afternoon a value of 11 % was observed. However, after midnight HP fell 31 % under the daily mean. The relation between daily minimum and maximum was 1.7:1. During night the subject standard deviation was 8.6% compared to about 12% in the daytime. It is concluded that the total experimental error observed generally in whole-body indirect calorimetry studies is mostly attributable to subject variability. Furthermore, the variation of energy expenditure may account for deviations up to 10% or more between individual requirement and recommendations for predicting energy require