The association between iron levels and coronary artery disease (CAD) mortality is controversial. Whereas most data show no association, some have raised the possibility of a causal role, while others have suggested a protective effect of iron on CAD. To address these possibilities, we examined the association between serum iron and CAD, cardiovascular disease, and all-cause mortality in a large cohort of 3,936 persons aged > or =71 years who completed an interview, had a serum iron determination, and survived at least 1 year after baseline. The median follow-up time was 4.4 years. Serum iron levels were categorized according to sex-specific quartiles. Relative risks (RR) and 95% confidence intervals (CI) were calculated from proportional-hazards regression models adjusted for age, race, education, creatinine, serum albumin, serum lipids, use of iron supplementation, smoking, use of alcohol, blood pressure, body mass index, and presence of chronic conditions. There was a gradual decrease in the RRs of CAD, cardiovascular disease, and all-cause mortality with increasing serum iron levels (all tests for trend, p <0.05). Men in the highest iron quartile were one fifth as likely to die of CAD as men in the lowest iron quartile (RR 0.22; 95% CI 0.11 to 0.48), and women in the highest quartile had half the risk of women in the lowest quartile (RR 0.48; 95% CI 0.27 to 0.87). When compared with the lowest quartile, risk of all-cause mortality was 38% lower in men in the highest iron quartile (RR 0.62; 95% CI 0.46 to 0.85) and 28% lower in women in the highest quartile (RR 0.72; 95% CI 0.53 to 0.96). Results of similar strength and magnitude were observed for cardiovascular disease mortality and in analyses that excluded the first 3 years of follow-up. In this large cohort of persons aged > or =71 years, there was consistent evidence of increasing risk of mortality at lower serum iron levels. In fact, lower serum iron levels were associated with an increased risk of CAD, cardiovascular disease, and all-cause mortality. The results are compatible with the possibility that in an older population, there is an inverse association between serum iron levels and risk of mortality.