The epidemiologic evidence linking high-density lipoprotein (HDL) levels with coronary artery disease (CAD) is persuasive. Case-control studies have shown CAD patients to have lower HDL levels than control subjects. Several large-scale, observational epidemiologic studies in the United States and abroad have shown a strong independent inverse relation between HDL and CAD. Women have a lower incidence of CAD than men of the same age; this has been attributed to their higher HDL levels. Postmenopausal women taking estrogen replacement therapy have higher HDL and lower low-density lipoprotein (LDL) levels, and a much lower incidence of CAD. Statistical analysis suggests that much of this is attributable to HDL levels. In several clinical trials, reduced levels of total or LDL cholesterol have been accompanied by increased HDL levels. Cox proportional hazards analysis suggests that the increment in HDL levels made an independent contribution to the reduction in CAD risk. In several angiographic studies, the increase in HDL may have contributed to the decreased progression, increased stabilization and possible regression of coronary lesions. Despite this range of impressive evidence, a number of unresolved issues have prevented the emergence of a consensus regarding the prevention of CAD by increasing HDL levels. Between-population comparisons of HDL and CAD do not match the within-population relations. Animal research on the relation between HDL, atherogenesis and CAD has been relatively scanty. Although much evidence suggests that reverse cholesterol transport partially explains the protective effect of HDL, there are still doubts as to its role. Problems with measurement of HDL have inhibited widespread recommendations for its use in prevention programs.(ABSTRACT TRUNCATED AT 250 WORDS)