Considerable epidemiological evidence has accumulated regarding the effect of postmenopausal estrogens on coronary heart disease risk. Five hospital-based case-control studies yielded inconsistent but generally null results; however, these are difficult to interpret due to the problems in selecting appropriate controls. Six population-based case-control studies found decreased relative risks among estrogen users, though only 1 was statistically significant. Three cross-sectional studies of women with or without stenosis on coronary angiography each showed markedly less atherosclerosis among estrogen users. Of 16 prospective studies, 15 found decreased relative risks, in most instances, statistically significant. The Framingham study alone observed an elevated risk, which was not statistically significant when angina was omitted. A reanalysis of the data showed a nonsignificant protective effect among younger women and a nonsignificant increase in risk among older women. Overall, the bulk of the evidence strongly supports a protective effect of estrogens that is unlikely to be explained by confounding factors. This benefit is consistent with the effect of estrogens on lipoprotein subfractions (decreasing low-density lipoprotein levels and elevating high-density lipoprotein levels). A quantitative overview of all studies taken together yielded a relative risk of 0.56 (95% confidence interval 0.50-0.61), and taking only the internally controlled prospective and angiographic studies, the relative risk was 0.50 (95% confidence interval 0.43-0.56).