Determinations of energy expenditure at rest were performed in 22 apparently healthy male subjects twice with one day's interval, and in 23 apparently healthy male subjects with one week's interval. In both series of experiments, the error of the method was calculated from multiple determinations on the same occasion. The coefficients of variation due to time were calculated and found to be 2.4 and 2.2 per cent, respectively. These numbers are practically identical to that found for the within-subject between-weeks variation in 24-h energy expenditure experiments using whole body calorimeters (Garby, Lammert & Nielsen, 1984).