One of the main goals of weight reduction in morbidly obese subjects is its benefit on coronary heart disease (CHD) risk. A cross-sectional study was designed to randomly assign 79 morbidly obese subjects (27 men and 52 women; age: 30-45 years) either to a diet protocol (20 kcal per kg fat-free mass (FFM); 55% carbohydrates, 30% fat, and 15% proteins) or to malabsorptive surgery (biliopancreatic diversion). Fatness parameters, measured by dual-energy X-ray absorptiometry, lipid profile, insulin, leptin, sex steroid hormones and sex hormone-binding globulin (SHBG) levels were compared at baseline and 1 year after the beginning of the study. The data showed that plasma SHBG levels, but not testosterone levels, correlated negatively to fasting insulin levels and positively to HDL-cholesterol in both men and women. Total leptin levels were significantly lower (P<0.0001) in post-BPD subjects of both sexes compared to dietary treated obese subjects. The logarithm of plasma leptin correlated significantly and positively with insulin but negatively with SHBG.A step-down regression analysis showed that FFM and SHBG, but not insulin levels, were the most powerful independent variables for predicting HDL-cholesterol levels in morbidly obese patients. The negative relationship between SHBG levels and CHD risk appears to be mediated by a concomitant variation in body fatness. Finally, in obese patients, SHBG levels seem to be an indicator of total adiposity rather than an index of an altered insulin/glucose homeostasis.