A technique is described for determining the apolipoprotein E genotype (apo E; alleles epsilon2, epsilon3, or epsilon4) from tissues which have been fixed with 4-10% formaldehyde and archived. The procedure requires efficient extraction and exhaustive purification of DNA from the fixed tissue. Because the fixation process renders the DNA largely crosslinked and/or sheared (therefore unsuitable for traditional analysis), a nested polymerase chain reaction (PCR) is employed (using two apo E gene specific primer pairs) to specifically amplify the polymorphic region of the gene. The genotype was then determined using previously reported HhaI polymorphisms that occur as a direct result of the variant codons responsible for the three alleles. This protocol permitted the successful genotyping of 90% (34 out of 38) of the archived brain samples from Alzheimer's disease (AD) patients. These samples included such extremes as a sample that had been stored for 12 years in formalin. This procedure permits the retrospective analysis of samples that had been processed and stored well before the original characterization of apo E alleles as risk factors in AD. Finally, this approach is readily adapted to the analysis of any gene of interest, whether by restriction fragment length polymorphism or direct amplicon DNA sequencing. It is also a very robust assay for less stringent conditions such as DNA isolated from whole blood or frozen tissue.