BACKGROUND AND PURPOSE: Both stroke and the apolipoprotein E (APOE) epsilon4 allele increase the risk of dementia. However, the interaction between stroke and APOE on dementia is still unclear. We addressed this topic by using a longitudinal design. METHODS: We followed up a community cohort of 1301 subjects aged >/=75 years, who did not have dementia at baseline. Among them, 92 subjects had a history of stroke (from 3 months to 16 years before baseline interview). After the 3-year follow-up, 224 dementia cases had been diagnosed. During the period of follow-up, 91 subjects had a first occurrence of stroke (incident stroke). The APOE genotype was known for 985 subjects. Cox proportional hazards regression models were constructed to estimate the risk for dementia in terms of relative risks (RRs) for stroke and the APOE epsilon4 allele, with adjustment for age, sex, education, systolic blood pressure, antihypertensive medication use, and heart disease. RESULTS: In the entire study population, RRs for dementia related to history of stroke and incident stroke were 1.7 (95% CI, 1.1 to 2.6) and 2.4 (95% CI, 1.6 to 3.5), respectively, after adjustment for all potential confounders. Subjects with stroke that occurred within 3 years before baseline had RR of 2.4 (95% CI, 1.4 to 4.2), whereas those with stroke occurring >3 years before baseline had RR of dementia of 1.1 (95% CI, 0.6 to 2.3). Among those with APOE information, individuals with only history of stroke (that occurred within 3 years before baseline) had RR of 3.1 (95% CI, 1.4 to 6.6), individuals with only the APOE epsilon4 allele had RR of 1.7 (95% CI, 1.1 to 2.5), and individuals with both factors had RR of 5.3 (95% CI, 2.1 to 13.4). The corresponding figures when incident stroke was examined instead of history of stroke were 2.3 (95% CI, 1.3 to 4.1), 1.7 (95% CI, 1.1 to 2.4), and 4.6 (95% CI, 2.0 to 10.6), respectively. The RR of interaction term for history of stroke and APOE epsilon4 was 1.1 (95% CI, 0.3 to 3.8; P=0.8). The corresponding figure was 1.2 (95% CI, 0.4 to 4.4; P=0.7) for incident stroke and APOE epsilon4. Furthermore, the RRs of dementia without any stroke and dementia with stroke in relation to APOE epsilon4 were 1.6 (95% CI, 1.1 to 2.3) and 1.2 (95% CI, 0.6 to 2.4), respectively. In addition, the APOE epsilon4 allele was not significantly related to the occurrence of stroke (RR=0.8; 95% CI, 0.5 to 1.5). CONCLUSIONS: A relatively fresh stroke is a risk factor for dementia. APOE epsilon4 increases the risk of dementia without stroke but not dementia with stroke. Our data do not support a multiplicative effect of stroke and the APOE epsilon4 allele on the risk of dementia. However, both factors seem to have an additive effect on the risk of dementia. The APOE epsilon4 allele does not increase the risk of stroke in this Swedish elderly population.