In adults, the high-density lipoprotein cholesterol (HDL-C) level is higher among physically active subjects. However, the association of physical activity and HDL-C is less well studied in adolescents and young adults. Furthermore, it is not known whether the effect of physical activity on HDL-C levels is independent, or whether it is mediated by other physiological changes seen in exercise, such as weight loss or increased insulin sensitivity. In order to study the effects of leisure-time physical activity on the levels of serum HDL-C concentration, we analysed longitudinal data from a follow-up study of adolescents and young adults. The study subjects were participants of a large multicentre study of cardiovascular risk factors, aged 15-21 years at the beginning of the study (n = 714). HDL-C was measured from the serum supernatant after precipitation with dextran sulphate and MgCl2. A physical activity index was calculated on the basis of frequency, intensity, and duration of leisure-time activity assessed by a questionnaire. In males, an increase in the physical activity level predicted an increase in HDL-C concentration, and this association persisted after simultaneously controlling for changes in body mass index (kg/m2), subscapular skinfold thickness, serum insulin and triglyceride concentrations, and smoking. For example, an increase in the physical activity level corresponding to approximately 1 hour of intensive exercise weekly lead to an increase of 42 mumol/L in HDL-C as calculated from the regression equation. In conclusion, physical activity seems to have a direct effect on HDL-C levels among young male subjects within the usual range of physical activity levels.