Body composition changes with increasing age in men, in that lean body mass decreases whereas fat mass increases. Whether this altered body composition is related to decreasing physical activity or to the known age-associated decrease in growth hormone secretion is uncertain. To address this question, three groups of healthy men (n = 14 in each group), matched for weight, height and body mass index, were investigated using dual-energy X-ray absorptiometry, indirect calorimetry and estimate of daily growth hormone secretion [i.e. plasma insulin-like growth factor I (IGF-I-) levels]. Group 1 comprised young untrained subjects aged 31.0 +/- 2.1 years (mean +/- SEM) taking no regular physical exercise; group 2 consisted of old untrained men aged 68.6 +/- 1.2 years; and group 3 consisted of healthy old men aged 67.4 +/- 1.2 years undergoing regular physical training for more than 10 years with a training distance of at least 30 km per week. Subjects in group 3 had for the past three years taken part in the 'Grand Prix of Berne', a 16.5-km race run at a speed of 4.7 +/- 0.6 min km-1 (most recent race). Fat mass was more than 4 kg higher in old untrained men (P < 0.01, ANOVA) than in the other groups (young untrained men, 12.0 +/- 0.9 kg; old untrained men, 16.1 +/- 1.0 kg; old trained men, 11.0 +/- 0.8 kg), whereas body fat distribution (i.e. the ratio of upper to lower body fat mass) was similar between the three groups. The lean mass of old untrained men was more than 3.5 kg lower (P < 0.02, ANOVA) than in the other two groups (young untrained men, 56.4 +/- 1.0 kg; old untrained men, 52.4 +/- 1.0 kg; old trained men, 56.0 +/- 1.0 kg), mostly because of a loss of skeletal muscle mass in the arms and legs (young untrained men, 24.0 +/- 0.5 kg; old untrained men 20.8 +/- 0.5 kg; old trained men, 23.6 +/- 0.7 kg; P < 0.01, ANOVA). Resting metabolic rate per kilogram lean mass decreased with increasing age independently of physical activity (r = -0.42, P < 0.005). Fuel metabolism was determined by indirect calorimetry at rest. Protein oxidation was similar in the three groups. Old untrained men had higher (P < 0.001) carbohydrate oxidation (young untrained men, 13.2 +/- 1.0 kcal kg-1 lean mass; old untrained men, 15.2 +/- 1.3 kcal Kg-1; old trained men, 7.8 +/- 0.8 kcal kg-1), but lower (P < 0.05, ANOVA) fat oxidation (young untrained men, 10.1 +/- 1.2 kcal kg-1 lean mass; old untrained men, 6.5 +/- 1.0 kcal kg-1; old trained men, 13.7 +/- 1.0 kcal kg-1) than the other two groups. Mean plasma IGF-I level in old trained men was higher than in old untrained men (P < 0.05), but was still lower than that observed in young untrained men (P < 0.005) (young untrained men, 236 +/- 24 ng mL-1; old untrained men, 119 +/- 13 ng mL-1; old trained men, 166 +/- 14 ng mL-1). In summary, regular physical training in older men seems to prevent the changes in body composition and fuel metabolism normally associated with ageing. Whether regular physical training in formerly untrained old subjects would result in similar changes awaits further study.