Variations in BMR, body weight and energy intake were measured for 14 consecutive days in 6 young adults on ad libitum energy intakes, whose physical activity was uncontrolled. Energy intakes showed significant differences between days (P less than 0.025, CV = 6.7 per cent), between weeks (P less than 0.005, CV = 8.9 per cent) and between subjects (P less than 0.005, CV = 7.9 per cent). Energy intakes were 14 per cent higher (P less than 0.01) at weekends. Intra-individual variance contributed up to 86 per cent of the total variance in the energy intake. Replicate BMR measurements showed non-significant differences from day to day (CV less than 1.5 per cent), a training effect from week to week (P less than 0.05, CV = 1.5 per cent), and significant differences between subjects (P less than 0.001, CV = 12.4 per cent). Intra-individual variance contributed only 14 per cent to the total variance in BMR. There were no significant changes in body weight (CV = 0.7 per cent) or fat-free mass during the study. Auto-correlations of BMR, body weight and energy intake were non-significant at the different lag times studied. Cross-correlations between the above parameters were also non-significant for each subject. It is concluded that despite wide fluctuations in energy intake from day to day within an individual, the variations in BMR are small with a true CV of less than 1.5 per cent. Hence these variations are unlikely to be important while assessing energy requirements on the basis of the FAO/WHO/UNU (1985) BMR factorial method.