In the Copenhagen Male Study, men with Lewis blood group phenotype Le(a- b-) were found to have increased risk for coronary heart disease (CHD); such a relation has not been confirmed in men, and has not been evaluated in women. In the NHLBI Family Heart Study, we determined the Lewis blood type of 1,620 white subjects (790 male and 830 female subjects). The Lewis(a-b-) phenotype was found in 142 subjects (8.8%), 6.3% of subjects from randomly chosen families and 9.7% of subjects from families found to be at high risk for CHD. A history of CHD was present in 39.1% of men with Le(a-b-) versus 27.2% of men with other Lewis types; for women, the corresponding numbers were 12.3% versus 9.4%, respectively. In multivariate analysis, adjusting for age, sex, and risk group, the odds ratio for CHD was 2.0 (95% confidence interval = 1.2 to 3.1) for Le(a-b-) versus other Lewis groups. Mean values for body mass index, blood pressure, total cholesterol, low-density lipoprotein and high-density lipoprotein cholesterol, glucose, insulin, homocysteine, and fibrinogen were not significantly different between Le(a- b-) subjects and others, but triglycerides (p = 0.002) were higher in the Le(a-b-) subjects. However, inclusion of all risk factors in multivariate analysis did not diminish the increased risk for CHD associated with the Le(a-b-) phenotype. We conclude that the Le(a-b-) phenotype is associated with an increased risk for CHD; its effect does not appear to act predominantly through conventional cardiovascular risk factors. At present, mechanisms of effect are unknown.