We used cross-sectional and exercise intervention studies to examine whether physical activity levels or increases in peak aerobic capacity (peak VO2) explain variation in high density lipoprotein cholesterol (HDL-C) levels in older men and women. In the cross-sectional study, 307 older individuals (169 men; 138 women; 67 +/- 7 years) were characterized for HDL-C, leisure time physical activity, peak VO2, body composition, body fat distribution and dietary intake. HDL-C was 19% higher (P < 0.001) in women (57 +/- 14 mg/dl) versus men (48 +/- 14 mg/dl). Thirty-two percent of the variation in HDL-C in older men was explained by the waist circumference (r2 = 16%), percent dietary intake of alcohol (r2 = 11%), and carbohydrate (r2 = 6%). Waist circumference was also the best predictor of HDL-C in older women, (r2 = 7%); with percent dietary intake of carbohydrate adding an additional 6% to the model. Neither peak VO2 nor leisure time physical activity were independent predictors of HDL-C. Statistical control for the aforementioned variables diminished, but did not abolish gender differences in HDL-C. Thirty-seven older individuals (23 men; 14 women) participated in a 2-month exercise program in which individuals by week eight were expending approximately 900 kcal per week in exercise energy expenditure. Subjects were maintained in energy balance throughout the exercise program. Endurance training significantly increased peak VO2 by 15% in both men and women, and by design, body composition and body fat distribution did not change. No changes in HDL-C levels were noted. In conclusion, variations in leisure time physical activity or increases in peak VO2 are not independent predictors of HDL-C levels in healthy older men and women. Instead, central adiposity, as estimated by the waist circumference, and to a lesser extent, dietary intake of carbohydrate and alcohol, are significant predictors of variation in plasma HDL-C levels. Furthermore, short-term exercise training, generating less than 900 kcal per week in exercise energy expenditure, in the absence of weight loss, fails to influence HDL-C levels.