The presence of apolipoprotein E-epsilon4 (APOE-epsilon4) allele has been implicated as a risk factor for Alzheimer's disease (AD). We examined the frequencies of APOE-epsilon4 alleles in age-matched controls and subgroups of 190 AD subjects exhibited cerebral amyloid angiopathy (CAA) and other frequently associated lesions. CAA was evident in 96% of the AD subjects, which were divided into two groups, one bearing mild or no apparent CAA and the other with moderate to severe CAA. APOE-epsilon4 allele frequency (48%) in the latter advanced CAA group was six times higher than in those who exhibited mild CAA. In the advanced CAA subjects, the occurrence of an epsilon4 allele was increased by a factor of 17 (95% confidence interval, 7.56 to 38.9). This was despite the fact that neocortical amyloid-beta plaque densities in the two groups were similar and that all of the AD subjects had met the accepted neuropathological criteria. We also observed that the degree of CAA severity was greatest in the group of subjects with the epsilon4/epsilon4 genotype. The association between CAA and APOE-epsilon4 was further implicated in two non-AD subjects among neurological controls with severe CAA. These two subjects, both homozygous for the APOE-epsilon4 allele, were primarily diagnosed as having Creutzfeldt-Jakob disease and Pick's disease in the absence of significant neocortical amyloid deposition. Allele frequency comparisons between neurological control subjects with CAA and those without likewise accorded a strong relationship between the APOE-epsilon4 allele and the presence of CAA. More remarkably, the epsilon4 allele frequency was highly associated with AD subjects exhibiting lobar or intracerebral hemorrhage, all of whom had advanced CAA. We observed that 36% of the AD subjects had concomitant cerebrovascular pathology resulting from single infarcts, multiple microinfarcts, ischemic white matter lesions, or petechial hemorrhages. Although the difference in APOE genotype distribution between subjects with and without cerebrovascular lesions did not reach statistical significance, we did note that the frequency of the epsilon4 allele was significantly higher in subjects with such pathology as compared with those without. However, we found no evidence to suggest that the acquisition of an APOE-epsilon4 allele or one of the alleles, epsilon2 or epsilon3, was a factor in the occurrence of atherosclerosis localized in the basal surface arteries. Analyses of our sample also confirm that there was a lower frequency of the APOE-epsilon2 allele in AD subjects and that the frequency of the epsilon4 allele in AD subjects with concomitant diffuse Lewy body disease was intermediate between controls and AD subjects. Our results suggest that the APOE-epsilon4 allele is a significant factor in the development of CAA in AD and reveal the possibility that APOE is an independent factor in CAA and other vascular abnormalities associated with AD.