Variability in the relationship of plasma leptin level to body mass index (BMI) could be caused by imperfect estimation of adipose mass by the BMI, heterogeneity in the pathogenesis of obesity in mixed subject groups, or variation in adipose tissue distribution. To investigate these possibilities, we examined the correlation of plasma leptin and BMI in an ethnically mixed population, a group of subjects with the Prader-Willi syndrome, and a group of Japanese-American subjects who underwent computerized tomographic measurement of adipose tissue cross-sectional areas. Highly significant and indistinguishable linear relationships between plasma leptin levels and BMI were found in the three study groups. Intersubject variability was also similar in the three groups and was reduced only when more accurate techniques for assessing adipose tissue mass were substituted for the BMI. The plasma leptin level of Japanese-American subjects in the highest quartile of intraabdominal fat area (mean area = 154.5 +/- 38.4 cm2) was 12.5 +/- 8.7 ng/mL as compared to 12.3 +/- 9.6 ng/mL (P = 0.91) for subjects in the lowest quartile of intraabdominal fat area (mean area = 51.2 +/- 20.1 cm2, P < 0.001 for difference in fat areas). We conclude that the circulating leptin level reflects total adipose tissue mass rather than a combination of adipose tissue mass and distribution, and that the Prader-Willi syndrome does not alter the relationship between these two variables.