Prospective data from the US Railroad Study were used to investigate the relations of several anthropometric variables to coronary heart disease and all-cause mortality in males initially free of cardiovascular disease. Middle-aged men were examined in 1957-1960 and followed until 1977 or death. Anthropometric indicators of total body fat (body mass index and the sum of the subscapular and triceps skinfolds) and central body fat (the ratios chest circumference/biacromial (shoulder) diameter, and chest circumference/standing height) were significantly and directly associated with age-adjusted coronary heart disease mortality. When systolic blood pressure, serum cholesterol, cigarette smoking, and vital capacity were also taken into account, the ratio chest circumference/biacromial diameter remained significantly associated with coronary heart disease mortality. Total, central, and peripheral body fat had a "U"-shaped relation with all-cause mortality. Measures of the ratio of central to peripheral body fat were inconsistently related to mortality. These results indicate that certain anthropometric measurements, especially those that may indicate central adiposity, are positively related to the development of fatal coronary heart disease and quadratically related to all-causes death rates.