The heart-rate (HR) method for determining the energy expenditure of free-living subjects has been evaluated using a whole-body calorimeter in which individuals lived continuously for 27 h while carrying out normal daily activities. Eight male volunteers each occupied the calorimeter on at least two occasions when HR and energy expenditure were measured continuously. 2. After each session in the calorimeter a calibration was obtained using standard techniques by determining HR and heat production (HP) over periods of 10-15 min at several levels of activity. Energy expenditure in the calorimeter was then predicted, by each of five methods, from the mean HR in the calorimeter. Additionally, one session in the calorimeter was used to obtain a calibration and was used for predicting the subject's energy expenditure while in the calorimeter on other occasions. 3. Standard methods of prediction using one calibration point at rest and several points during activity were unreliable for predicting the energy expenditure of an individual. The 24 h HR was at the lower end of the calibration scale and there were considerable over-estimates or underestimates of energy expenditure, particularly during the night when the mean (+/- SD) difference between the actual and predicted HP was -66 +/- 38.6%. A linear regression fitted to points at the lower levels of activity improved the prediction of 24 h HP while a logistic plot reduced the error even further. The best estimate of energy expenditure was that obtained from a calibration over 24 h within the calorimeter; the mean (+/- SD) difference between the actual and predicted 24 h HP was +3 +/- 10.5% for light activity and -3 +/- 6.7% for moderate activity. Thus current procedures for calibrating subjects may lead to large errors which could be reduced by using a respiratory chamber.
