Estimates of the impact of apolipoprotein E (apo E) alleles coding for the three common isoforms on plasma lipid levels assume genetic homogeneity among the genotype classes. To test this assumption, we have determined the apo E genotype at the two common polymorphic sites (amino acids 112 and 158) by DNA amplification and hybridisation with allele-specific oligoprobes, in 195 unrelated Caucasian participants of the Rochester Family Heart Study previously classified as heterozygotes by isoelectric focusing (IEF). Fourteen discordant samples were initially detected. Repeat typing of these samples by both methods resolved nine discrepancies and analysis of additional blood samples from the remaining five individuals eliminated a further four discrepancies. The only truly discordant allele was found in a female subject who had an E3 isoform with the common E2 (CYS,,~, C Y S ~~~) genotype. Transmission of this allele from the mother was demonstrated. From these results, we estimate the frequency of discrepancies between isoforms and common genotypes to be 0.25% in this population. Allele misclassification was caused by poor amplification of the DNA in six samples and superimposition of glycosylated and nonglycosylated apo E isoforms on isoelectric focusing gels in five samples. We conclude that the assumption of genetic homogeneity among genotype classes is valid and that misclassification due to technical difficulties is more frequent than true discordancies. o 1992 Wiley-Liss, Inc.