The existence of a set-point for homeostatic control of human body weight is uncertain. To investigate its existence, technically difficult determinations of energy expenditure must be performed: this has resulted in contradictory reports. The present study was performed with new methods in two stages (77 and 133 d respectively). Two healthy male subjects with rigorously controlled physical activity ingested three standardized diets of processed foods from the same manufacturer. Hypo-, iso- and hyperenergetic diets containing 6255 kJ (1494 kcal), 10 073 kJ (2406 kcal) and 13 791 kJ (3294 kcal) respectively were ingested during alternate periods; changes in body weight were measured. A new index of energy expenditure was calculated as the amount of weight lost in an 8 h overnight period (WL8H). A digital scale was used in stage 1 and a mechanical scale in stage 2. The change in body weight in response to the isoenergetic diet differed according to the circumstances. In basal conditions, it was associated with weight stability. After weight loss from energy restriction, the isoenergetic diet led to weight gain. After weight gain from overeating, it led to weight loss. Diets of higher energy content were associated with greater WL8H (F>20; P<0.0001 for both subjects). Measurement variability was lower using a mechanical scale. The present study demonstrates the existence of a homeostatic control of human weight and describes a new index of energy expenditure measured in weight units. It also demonstrates that strict dietary supervision for months is possible. Investigation of the human body weight set-point is vital in understanding obesity.