Apolipoprotein E (apo E) genotypes have been determined in 460 Danish men, with the use of the polymerase chain reaction (PCR) to amplify a 244 base pair fragment spanning the first-base polymorphic sites in the codons of amino acids 112 and 158 followed by restriction endonuclease cleavage. The results were compared with the apo E phenotypes previously determined by isoelectric focusing (IEF) of delipidated plasma, not pretreated with neuraminidase, followed by apo E specific immunoblotting. Conflicting results were found in only 9 cases (2.0%) and in each case only with respect to one allele. Five of the discrepancies can be explained, post hoc, by technical difficulties with the IEF method ('faint bands'). A possible cause of the other 4 discrepancies is the presence of rare mutations. Our findings in this large study are reassuring, since, if appreciable and systematic misclassification of genotypes do occur by using IEF, as has been reported from some laboratories, it may influence the validity of genetic epidemiological studies.