The distribution of serum concentrations of HDL-cholesterol (HDL-C), its correlation with other risk factors, its determinants and its association with coronary heart disease (CHD) likelihood was studied cross-sectionally in 2350 persons constituting the 2001/02 cohort of the Turkish Adult Risk Factor Study. Moreover, the prediction of future fatal and nonfatal CHD was investigated in the 1997/98 cohort comprising 2362 men and women free of CHD. HDL-C was generally measured with dry chemistry method using a Reflotron apparatus, but utilizing the direct method without precipitation in year 2001 and was validated in over 6% of the cohort in a reference laboratory. Mean HDL-C was 38.0±10.0 and 45.3±11.3 mg/dl in men and women, respectively, whose mean age was 52±12 years in the cross-sectional analysis. HDL-C levels were under 40 mg/dl in 64% of men and in 35.5 of women. The inverse relationship between HDL-C and triglyceride concentrations was demonstrated in increasing deciles. In linear regression models comprising 11 variables, smoking habit, waist circumference and physical inactivity suppressed HDL-C, whereas alcohol use was associated with elevation of this level. The slope of the linear regression allowed following estimated independent changes (p<0.01) in HDL-C corresponding to an increment of 1 standard deviation: 80 mg/dl triglycerides 1.6 mg/dl decline in HDL-C, 12 cm waist circumference 1.33 mg/dl decline in HDL-C, smoking more than 10 cigarettes daily 3.16 mg/dl decline in HDL-C, apo B 40 mg/dl 2.76 mg/dl decline in HDL-C, 40 mg/dl total cholesterol 3.3 mg/dl rise in HDL-C. Alcohol use once a week or more frequently was associated with a rise of 3.5 mg/dl of HDL-C (p=0.064). Over a 4-year prospective study, logistic regression analysis comprising 10 variables including systolic blood pressure, smoking habit, total cholesterol and diabetes, revealed HDL-C to be an independent predictor of CHD with a relative risk of 0.971 in men (p<0.02) and of 0.980 in women (p=0.08). Among men and women combined, RR was 0.975 (p<0.002), which implied that a reduction of HDL-C by 12 mg/dl (=1 hazard ratio), would be associated with 36% increase of fatal and nonfatal events. The relative roles of main biochemical and lifestyle determinants of HDL-C levels in Turkish adults were thus established. The major dimensions of effects of obesity with associated hypertriglyceridemia, excessive smoking and abstinence of alcohol on HDL-C levels are apparent. HDL-C concentrations represent a significant independent predictor of future coronary events; however, its magnitude does not appear to be very strong.