Although apolipoprotein E epsilon4 is an established risk factor for Alzheimer's disease, its effect on the rate of progression of Alzheimer's disease remains unknown. The purpose of this longitudinal study was to elucidate whether the rate of hippocampal atrophy is a function of the apolipoprotein E genotypes and severity of disease. Fifty-five patients with probable Alzheimer's disease were the subjects. The annual rate of hippocampal atrophy was determined by using magnetic resonance imaging repeated at a 1-year interval. On a two-way analysis of variance, the effect of the apolipoprotein E epsilon4 allele on hippocampal atrophy was significant, but neither the effect of severity nor the interaction term was significant. In further analysis with one-way analysis of variance, the mean annual rate of hippocampal atrophy was significantly different between the groups of patients with (9.76 +/- 4.27%) and without the apolipoprotein E epsilon4 allele (6.99 +/- 4.24%). Apolipoprotein E epsilon4 dose was significantly correlated with the rate of hippocampal atrophy (rs = 0.277, Spearman rank correlation coefficient), suggesting a gene dose effect. The involvement of the apolipoprotein E epsilon4 allele in the progression of hippocampal atrophy has implications for therapeutic approaches in Alzheimer's disease and should be taken into consideration in longitudinal studies including clinical drug trials.