There is a limited knowledge concerning energy requirements of the elderly, especially the oldest old (> 80 years). Energy requirements should be estimated from measurements of energy expenditure. For this purpose twenty-one free-living individuals (eight males, thirteen females) aged 91–96 years living in Göteborg, Sweden were studied. Total body water (TBW) measured by the doubly-labelled-water (DLW) technique was 29·5 (SD 5·4) KG IN FEMALES AND 35·6 (sd 4·3) kg in males. TBW measured using bioelectric impedance (BIA) was 31·6 (sd 6·4) kg in females and 42·0 (sd 7·4) kg in males. The mean difference between TBW measured by BIA and that measured by DLW was 3·54 (sd 3·6) kg (P = 0·0002). Resting metabolic rate (RMR) was measured using a ventilated-hood system and averaged 5·36 (sd 0·71) MJ/d in females (n 12) and 6·09 (sd 0·91) MJ/d in males (n 8). Difference between measured RMR and predicted BMR (n 20) was 0·015 (sd 0·86) MJ/d (NS). Total energy expenditure (TEE) measured by DLW averaged 6·3 (sd 0·81) MJ/d in females and 8·1 (sd 0·73) MJ/d in males. Activity energy expenditure (TEE - RMR), thus including diet-induced thermogenesis (DIT), averaged 0·95 (sd 0·95) MJ/d in females (n 12) and 2·02 (sd 1·13) MJ/d in males. Physical activity level (TEE/BMR) averaged 1·19 (sd 0·19) in females and 1·36 (sd 0·21) (P = 0·08) in males. If DIT is assumed to be 10 % of the TEE, energy spent on physical activity will be very low in this population.