Excess visceral adipose tissue (AT) and hyperinsulinemia are important correlates of an altered lipoprotein profile. It has also been reported that testosterone, adrenal C19 steroids and sex hormone-binding globulin (SHBG) concentrations are associated with plasma lipoprotein levels. The aim of the present study was to investigate the relative contributions of endogenous steroid hormone and SHBG levels, of visceral AT accumulation measured by computed tomography, and of fasting insulin and free fatty acid (FFA) concentrations to the variation of plasma lipoprotein levels in men. For this purpose, plasma concentrations of testosterone, dehydroepiandrosterone (DHEA), androstene-3beta,17beta-diol (delta5-DIOL), androstenedione (delta4-DIONE), estrone and estradiol, as well as SHBG levels were determined in a sample of 76 men covering a wide range of body fatness values. Higher testosterone levels were associated with a more favorable lipoprotein profile as it showed significant correlations with triglyceride (TG), total cholesterol and LDL-cholesterol (LDL-C) concentrations (r= -0.25, -0.25 and -0.27, respectively; P < 0.05). Higher plasma adrenal C19 steroid levels were also associated with a favorable lipoprotein profile as DHEA, delta4-DIONE and delta5-DIOL levels were negatively correlated with total cholesterol (r = -0.24, -0.33 and -0.24, respectively; P < 0.05) and LDL-C (r = -0.23, -0.31 and -0.28, respectively; P < 0.05). SHBG levels were negatively correlated with TG concentrations (r = -0.33; P < 0.005) whereas delta5-DIOL, testosterone and SHBG were negatively correlated with apolipoprotein B levels (-0.32 < or = r < or = -0.43; P < 0.005). Statistical adjustment for visceral AT area, fasting insulin, fasting free fatty acid (FFA) levels and total body fat mass eliminated most of the correlations between steroid and lipoprotein levels, while SHBG remained significantly correlated with lipoprotein concentrations after such adjustments. Multivariate analyses revealed that SHBG, delta4-DIONE, delta5-DIOL and metabolic variables all contributed to the variance in plasma lipoprotein concentrations (from 10 to 29% of explained variance). Visceral AT, fasting FFA and insulin levels as well as SHBG concentrations appeared to be independent correlates of lipoprotein concentrations. Thus, metabolic and anthropometric variables examined in the present study could have represented important confounding factors in previous studies which have examined the relationship of steroid hormones to plasma lipoprotein concentrations.