Although the APOE epsilon4 allele is a strong risk factor for Alzheimer's disease (AD), it is not deterministic, as many APOE epsilon3/4 individuals do not develop AD. It has been hypothesized that this incomplete penetrance is due, in part, to an imbalance of allele expression in heterozygous individuals. In this regard, Lambert et al. (1998) reported that AD individuals have a higher APOE epsilon4/total APOE ratio than non-demented control subjects. We tested this hypothesis using radioactive RT-PCR to quantitate APOE epsilon3 and epsilon4 allele expression levels in AD and non-AD brain samples from APOE epsilon3/4 individuals. Quantitative analyses of amplified products within the linear range of amplification (18-20 cycles) revealed no difference from the expected 1:1 ratio in genomic DNA and in cDNA from AD and control brains. Using high PCR cycle numbers (approximately 30), we observed an artificial elevation of the APOE epsilon3/total APOE ratio in both DNA and cDNA samples, possibly due to DNA heteroduplex formation. Our results do not support the hypothesis that allelic imbalance contributes to the risk of developing AD among APOE epsilon3/4 heterozygote individuals.