The doubly-labelled water (2H218O) technique was used to assess the long-term rates of energy expenditure and, after accounting for any changes in body composition, the derived rates of energy intake in weight-stable ‘large-eating’ (n 6) and ‘small-eating’ (n 6) women. The self-reported energy intakes (approximately 11.2 v. 5.6 MJ/d) and energy expenditures (approximately 8.5 v. 12.4 MJ/d) for the ‘large-eating’ and ‘small-eating’ groups respectively, should not be sustainable without significant body-weight changes. 2H218O-assessed rates of energy expenditure for the ‘large-eaters’ (approximately 8.5 MJ/d) and ‘small-eaters’ (approximately 11.3 MJ/d) were in close agreement with the results obtained using 5 d, self-reported activity diaries but the derived rates of energy intake for the ‘large-’ (approximately 8.5 MJ/d) and ‘small-eaters’ (approximately 10.8 MJ/d) were markedly different from those obtained using self-reported, weighed food diaries. When two ‘small-eaters’ were supplied with their self-reported energy intakes (approximately 5 MJ/d) for up to 28 d both subjects lost about 0.75 kg body-weight/week. These results provide no support for the existence of ‘metabolically efficient’ women in the community.