In a recent study, we could account for only about 50% of the variance in angiographically determined severity of coronary artery disease (CAD) with use of lipid and clinical variables as predictors. To explore the possible contribution of the apolipoprotein (apo) E polymorphisms to the severity of CAD (rather than to its occurrence), we studied 424 white patients aged 65 years or less consecutively referred for coronary angiography. Among the 304 male and 120 female patients, there were 110 with no significant CAD and 118 with one, 96 with two, and 100 with three significantly diseased major coronary arteries (> 50% luminal obstruction). The allele frequencies were 0.068 for E2, 0.759 for E3, and 0.172 for E4. The E2 frequency was slightly lower and E4 higher than the frequencies reported for healthy white populations (E2: 0.072 to 0.130; E4: 0.136 to 0.160). There was a clear association between the apo E genotype and the number of significantly diseased vessels (regression coefficient = .12, P = .008). The frequency of the E4 allele increased linearly with the increase in CAD severity in both sexes (for none, one, two, and three significantly diseased vessels; female patients: 0.136, 0.161, 0.200, and 0.324; male patients: 0.136, 0.167, 0.132, and 0.229, respectively, P < .01). The frequencies of the E2 allele, on the other hand, decreased with increasing severity (for none, one, two, and three significantly diseased vessels; female patients: 0.091, 0.018, 0.050, and 0.029; male patients: 0.073, 0.089, 0.072, and 0.054, respectively, P < .05).(ABSTRACT TRUNCATED AT 250 WORDS)