A prior history of depression and the epsilon 4 allele of apolipoprotein E (APOE) have each been associated with development of Alzheimer's disease (AD). In a sample of 142 elderly twins from a large study of dementia, we examined the relation of major depression, APOE genotype and AD using time-dependent proportional hazards models. Compared against the risk for AD with no history of depression and no epsilon 4 allele, the risk ratio for AD with two epsilon 4 alleles was 2.87 (C.I. = 1.56-5.28), with one epsilon 4 allele, 1.82 (C.I. = 1.09-3.04) and with late-onset depression and no epsilon 4 allele, 2.95 (C.I. = 1.55-5.62). There was no suggestion of an interaction between prior depression and APOE genotype in their effects on AD risk. Results were similar when the sample was stratified by twin pair, so that a single genetic marker is unlikely to explain the relation among depression, APOE, and dementia. Risk ratios declined substantially with increasing intervals between the onset of depression and AD. Thus, for many individuals, the association of depression and AD may reflect the occurrence of prodromal depressive symptoms rather than a true risk relationship.