We determined the contribution of body fat distribution, peak VO2, fat mass, and dietary intake to variation in plasma lipids in elderly individuals. Volunteers were a healthy cohort of older Caucasian women (n = 75, mean age +/- SD, 72 +/- 5 years) and older men (n = 101, 72 +/- 5 years). We determined fat mass from underwater weighing, fat patterning from waist circumference, as well as peripheral and truncal skinfolds, exercise capacity from peak VO2, and dietary intake from three-day food diaries. Plasma lipid levels were measured in the fasting state and included total cholesterol, high density lipoprotein (HDL-C), low density lipoprotein (LDL-C), and fasting triglycerides. Older women weighted less than older men, but had higher fat mass, truncal, and peripheral skinfolds. Waist circumference and peak VO2 were lower in older women than older men. Older women had higher total cholesterol (217 +/- 31 vs. 197 +/- 30; p < 0.01), HDL-C (54 +/- 12 vs. 49 +/- 14; p < 0.05), and LDL-C (133 +/- 26 vs. 121 +/- 27; p < 0.01) when compared with older men. No gender differences were noted in fasting triglycerides. Truncal skinfolds were the best predictor of plasma lipids in older men, accounting for between 9% and 30% (r2) of the variation in plasma lipids. Similarly, in older women, central markers of fatness (i.e., waist circumference and truncal skinfolds) were the best predictors of plasma lipids (r2 = 3% to 24%). Total fat mass, peak VO2 and dietary intake were not independent predictors of plasma lipids in older men and women. Indices of central body fatness, rather than total fat mass, peak VO2 or dietary intake are stronger predictors of plasma lipids in healthy older men and women.