BACKGROUND: The iron hypothesis suggests that females are protected from atherosclerosis by having lower iron stores than men, thus limiting oxidation of lipids. OBJECTIVE: To test the iron hypothesis by comparing cardiovascular event rates in whole blood donors compared with nondonors. DESIGN: Prospective cohort with telephone survey follow up. SETTING: The State of Nebraska, USA. PARTICIPANTS: A sample was selected from the Nebraska Diet Heart Survey (NDHS) restricting for age > or = 40 years and absence of clinically apparent vascular diseases at time of enrollment in to NDHS (1985-87). MAIN OUTCOME MEASURES: The occurrence of cardiovascular events (myocardial infarction, angina, stroke), procedures (angioplasty, bypass surgery, claudication, endarterectomy), nitroglycerin use, or death (all cause mortality), and level of blood donation. RESULTS: Participants were 655 blood donors and 3200 non-donors who differed in education, physical activity, diabetes, and frequency of antihypertensive treatment; 889 were lost to follow up. Sixty four donors and 567 non-donors reported cardiovascular events (crude odds ratio = 0.50, 95% confidence interval (CI) 0.38-0.66). The benefit of donation was confined to non-smoking males (adjusted odds ratio 0.67, 95% CI 0.45-0.99). Benefit was limited to current donors (the most recent three years). No additional benefit resulted from donating more than once or twice over three years. CONCLUSION: In support of the iron hypothesis, blood donation in non-smoking men in this cohort was associated with reduced risk of cardiovascular events. A randomised clinical trial is warranted to confirm these findings as the observed personal health benefit of donation has public policy ramifications.