Background The epsilon 4 allele of the apolipoprotein E gene ( APOE4 ) plays a role in neurodegeneration and in cardiovascular disease, but findings on its association with mortality are inconsistent. We aimed to examine the association between APOE4 and mortality, and the role of dementia in this association. Methods In this pooled analysis, data on White participants aged 45-90 years who underwent APOE genotyping were drawn from two population-based cohorts: the Whitehall II study (UK), which began in 1985 and is ongoing, and the Three-City study (France), initiated in 1999 and ended in 2012. In the Three-City study, vital status was ascertained through linkage to the national registry of death Institut National de la Statistique des Etudes Economiques, and dementia was ascertained via a neuropsychological evaluation and validation of diagnoses by an independent committee of neurologists and geriatricians. In the Whitehall II study, vital status was ascertained through linkage to the UK national mortality register, and dementia cases were ascertained by linkage to three national registers. Participants with prevalent dementia at baseline and participants missing an APOE genotype were excluded from analyses. Cox regression proportional hazard models were used to examine the association of APOE4 with all-cause, cardiovascular, and cancer mortality. The role of dementia in the association between APOE4 status and mortality was examined by excluding participants who developed dementia during follow-up from the analyses. An illness-death model was then used to examine the role of incident dementia in these associations. Findings 14 091 participants (8492 from the Three-City study and 5599 from the Whitehall II study; 6668 [47%] of participants were women and 7423 [53%] were men), with a median follow-up of 154 years (IQR 106-212), were included in the analyses. Of these participants, APOE4 carriers (3264 [23%] of the cohort carried at least one epsilon 4 allele) had a higher risk of all-cause mortality compared with non-carriers, with hazard ratios (HR) of 116 (95% CI 107-126) for heterozygotes and 159 (124-206) for homozygotes. Compared with APOE3 homozygotes, higher cardiovascular mortality was observed in APOE4 carriers, with a HR of 123 (101-150) for heterozygotes, and no association was found between APOE4 and cancer mortality. Excluding cases of incident dementia over the follow-up resulted in attenuated associations with mortality in homozygotes but not in heterozygotes. The illness-death model indicated that the higher mortality risk in APOE4 carriers was not solely attributable to dementia. Interpretation We found a robust association between APOE4 and all-cause and cardiovascular mortality but not cancer mortality. Dementia explained a significant proportion of the association with all-cause mortality for APOE4 homozygotes, while non-dementia factors, such as cardiovascular disease mortality, are likely to play a role in shaping mortality outcomes in APOE4 heterozygotes. Copyright (c) 2024 The Author(s). Published by Elsevier Ltd.