Numerous interrelated metabolic and morphological variables such as plasma insulin levels, glucose tolerance and abdominal obesity are associated with changes in plasma lipoprotein levels. The present study was undertaken to differentiate, using a multivariate approach, the respective contributions of plasma glucose and insulin levels, obesity and regional adipose tissue distribution to the variance in plasma lipoproteins. The study group was composed of 69 healthy premenopausal women (age 35.4 +/- 5.0 years (mean +/- s.d.); percent body fat 40.7 +/- 10.1). Indices of carbohydrate metabolism showed significant univariate correlations with triglyceride (TG) and/or cholesterol (CHOL) content of plasma VLDL, LDL and HDL (P less than 0.05). Multivariate analyses indicated that the explained variance in plasma VLDL-TG (R2 x 100 = 44 percent, P less than 0.05) and LDL-apoprotein (apo) B levels (R2 x 100 = 33.1 percent, P less than 0.08) was entirely accounted for by indices of carbohydrate metabolism and body fat distribution, whereas total body fatness added no significant contribution to these models. Multivariate analyses also revealed that the best possible regression model to predict the variation in plasma HDL2-CHOL levels only included computed tomography-derived deep abdominal adipose tissue area (P less than 0.0001). All other variables were unable to further improve the explained variance in plasma HDL2-CHOL levels. In partial correlation analyses, indices of carbohydrate metabolism and the waist-to-hip circumference ratio (WHR) remained significantly correlated with plasma VLDL-TG and LDL-apo B levels after adjustment of VLDL-TG and LDL-apo B for either insulin and glucose levels, or for the WHR (P less than 0.08). After correcting for deep abdominal fat accumulation, no significant correlation was observed between indices of carbohydrate metabolism and plasma HDL2-CHOL levels whereas deep abdominal fat showed significant correlations with HDL2-CHOL levels (P less than 0.05) after correction for indices of carbohydrate metabolism. These results suggest that both disturbances in glucose-insulin homeostasis and abdominal obesity are significantly associated with changes in plasma VLDL-TG and LDL-apo B levels and that these associations are partly independent from each other. These results also indicate that mechanisms other than disturbances in glucose homeostasis and hyperinsulinemia are responsible for the association between the level of deep abdominal fat and plasma HDL2-CHOL levels.