CONTEXT: Although endogenous estradiol, generally considered as the female hormone, has been little investigated in men, it could play a role in men's health, mortality in particular. The influence of estrogen receptors (ER) genetic polymorphisms on this relationship has never been studied. DESIGN AND PARTICIPANTS: The Three-City cohort study included (1999-2001) 3650 men >/=65 years who were followed for mortality over 12 years. At baseline, total estradiol (tE2) was measured and ER genotyped in a random subsample of 472 men without hormonal treatment. Free estradiol (fE2) was estimated using Vermeulen and Sodergard algorithms. MAIN OUTCOME: Mortality data were obtained from death certificates. We used inverse probability weighted Cox models to examine the association of estradiol with all-cause and cause-specific mortality and its interaction with ER genetic polymorphisms. RESULTS: 183 men died over the follow-up (cardiovascular disease (CVD), n=44; cancer, n=57; other causes, n=82). After adjustment, there was a quadratic relationship of all-cause mortality with tE2 and fE2 (p-quadratic=0.04 and 0.05, respectively), with higher mortality for the top and bottom tertiles compared to the middle one. These associations were stronger for CVD mortality (p-quadratic=0.01 and 0.02 for tE2 and fE2, respectively) and disappeared for cancer mortality. There was no evidence of an interaction of estradiol with any ER polymorphisms on all-cause mortality. CONCLUSION: In elderly men, we showed a nonlinear association of tE2 and fE2 with all-cause mortality. These quadratic relationships were stronger for CVD mortality and did not exist for cancer mortality. ER genetic polymorphisms did not modify this association. This article is protected by copyright. All rights reserved.< Réduire