Eight Friesian calves were reared from birth to the start of the experiment at a predetermined growth rate of 1.0 kg/d. Four calves (group 1) commenced the experiment at 22 weeks of age and the remainder (group 2) at 34 weeks, and feed intake was calculated to provide sufficient metabolizable energy to sustain zero energy balance (Em). Between experimental days 30 and 56 at maintenance levels of feeding, respiratory exchange measurements were made using open-circuit calorimetry to provide indirect estimates of heat production (H) simultaneous with measurements of faecal and urinary excretion of energy and N. A datum point for Em and H was established for each animal giving mean values for Em of 491 and 537 and for H of 476 and 511 kJ/kg live weight (LW)0.75 per d for groups 1 and 2 respectively. Treatment levels calculated as 0·25, 0·5, 0·75 and 1·25 of dry matter intake required to sustain the measured Em, were imposed on individual animals after day 56 and further measurements of H and faecal/ urinary balance were made between days 72 and 84 followed by measurements of fasting heat production (FHP) during days 86–90 from the two animals in each group which had received the 0.75 and 1.25 levels of intake. Regression analysis of the treatment levels indicated separate linear models which predicted Em at 419 and 473 kJ/kg LW0·75 per d for groups 1 and 2 respectively. The incorporation of FHP with partitioning of faecal and urinary energy losses measured during fasting altered the relationship but not the predicted Em. Overall predicted Em (days 72–84) from all models were 406 and 478 kJ/kg LW0.75 per d for groups 1 and 2 respectively which were significantly lower (P < 0·05) than Em measured during days 30–56.