Correlations of high-density lipoprotein cholesterol (HDL-C), expressed in either absolute or relative terms, with a series of coronary risk factors and other variables were examined in the Israeli Ischemic Heart Disease Study sample. The Quetelet overweight index showed the highest correlations with HDL-C (r = -0.21) as well as with HDL-C as a percentage of total cholesterol (TC) (HDL/TC; r = -0.28). Additional negative inverse associations were statistically different from zero but small. High-density lipoprotein cholesterol and HDL/TC were significantly reduced in cigarette smokers, and HDL/TC was significantly reduced in men with myocardial infarction or angina pectoris and (albeit marginally) in diabetes mellitus as well. The presence of these diseases correlated poorly with HDL-C (absolute values). The associations of HDL-C and HDL/TC with the Quetelet index persisted after adjustment for cigarette smoking and vice versa. Reported dietary intake failed to explain HDL-C or HDL/TC variability among individuals. The "net" relationship of HDL-C to several variables was examined in a multiple regression analysis. The Quetelet index accounted for 0.21 of a multiple correlation coefficient of 0.28 (i.e., a very small proportion of explained variability). This magnitude is of an order similar to multiple correlations found in our study for total cholesterol, systolic blood pressure, and serum uric acid. It indicates that our knowledge of the determinants of HDL-C in adults is insufficient. The possible roles of several anthropometric and behavorial variables in determining HDL-C levels are considered, as is the possible genetic factor in dictating interindividual HDL-C variability.