The objective of the present study was to investigate the contribution of intra-individual variance of resting energy expenditure (REE) to interindividual variance in REE. REE was measured longitudinally in a sample of twenty-three healthy men using indirect calorimetry. Over a period of 2 months, two consecutive measurements were done in the whole group. In subgroups of seventeen and eleven subjects, three and four consecutive measurements were performed over a period of 6 months. Data analysis followed a standard protocol considering the last 15min of each measurement period and alternatively an optimised protocol with strict inclusion criteria. Intra-individual variance in REE and body composition measurements (CVintra) as well as interindividual variance (CVinter) were calculated and compared with each other as well as with REE prediction from a population-specific formula. Mean CVintra for measured REE and fat-free mass (FFM) ranged from 5·0 to 5·6% and from 1·3 to 1·6%, respectively. CVintra did not change with the number of repeated measurements or the type of protocol (standard v. optimised protocol). CVinter for REE and REE adjusted for FFM (REEadj) ranged from 12·1 to 16·1% and from 10·4 to 13·6%, respectively. We calculated total error to be 8%. Variance in body composition (CVintra FFM) explains 19% of the variability in REEadj, whereas the remaining 81% is explained by the variability of the metabolic rate (CVintra REE). We conclude that CVintra of REE measurements was neither influenced by type of protocol for data analysis nor by the number of repeated measurements. About 20% of the variance in REEadj is explained by variance in body composition.