We studied 24 healthy men (25-50 years old) covering a wide range of fatness (body mass index range: 21-34 kg/m2) and fat distribution (waist/hip range: 0.75-1.06). Computed tomography scans were taken at five levels (thigh, hip, waist, arm, and liver) from which fat, muscle and bone areas were calculated. Both waist/hip and BMI were correlated with fat areas in the thigh, arm and waist scans. BMI showed stronger correlations with peripheral fat areas, whereas waist/hip showed stronger correlations with fat areas in the waist scan (particularly with visceral fat area: r = 0.88, P less than 0.001). BMI was correlated with muscle and bone areas in the thigh scan. In multiple regression BMI was, independently of waist/hip and age, positively correlated with fat areas in the arm, thigh, and waist (not with visceral fat) and muscle and bone areas in the thigh. Waist/hip was independently of BMI and age correlated with fat areas in the arm and waist, including visceral fat area (but not with fat areas in the thigh). Moreover, waist/hip showed an independent negative correlation with muscle area in the thigh, muscle endurance and physical activity. Serum triglycerides, plasma insulin, glucose, uric acid and diastolic and systolic blood pressure were associated with visceral fat area but also to anthropometric indicators of abdominal fat distribution (especially waist/hip ratio). Liver attenuation, but not the liver/spleen attenuation ratio, was associated with some liver enzymes and BMI but not with waist/hip or metabolic parameters. We conclude that a higher BMI is associated with increased central and peripheral fat stores (but not visceral fat) and increased thigh muscle whereas waist/hip is primarily associated with increased central fat stores (noteably with visceral fat), decreased thigh muscle and reduced physical fitness. It is suggested that physical training might be an important element in the treatment of abdominal obesity in men.