A study was conducted to investigate the validity of skinfold-based prediction equations for body density (g/ml) by Nagamine and Suzuki (1964), and to formulate more convenient and more useful equations for predicting body density from skinfold and age in men. Subjects of the study were 257 healthy men aged 19-60 years in or near Nagasaki City. The regression equation for the dependent variable, body density, was determined by hydrostatic weighing. Independent variables included eight skinfolds, the sum of two skinfolds (triceps, subscapular), the sum of three skinfolds (triceps, subscapular, and abdominal), age, and body surface area. Skinfolds were measured with an Eiken-model skinfold caliper. Age (mean 33.1, range 19-60 yrs.), weight (mean 65.3, range 46.6-107.7 kg), height (mean 168.8, range 152.3-185.4 cm), and body density (mean 1.05874, range 1.00860-1.09020 g/ml) were also recorded. Percent body fat was calculated using the formula by Brozek et al. and ranged from 6.1% to 38.9%. Multiple correlation coefficients (MR) and standard error (SE) of 10 regression equations (A-J) for predicting body density in men were obtained. The best-fitting and the most convenient prediction equation for body density was equation-E.: body density = 1.09556-0.00062 x sum of three skinfolds (mm)-0.00028 x age (MR = 0.815 and SE = 0.0087 g/ml). The equation was cross-validated on a different sample of 45 men. The correlation coefficient between predicted and hydrostatically determined body density was 0.781 (p < 0.001). Equation-E (Tahara's equation) appears to be useful in body density analysis particularly when the subjects are Japanese men, aged 18-50 yrs, with percent body fat 10 to 30%.