The association between serum ferritin level and coronary heart disease (CHD) and stroke events was evaluated in a long-term Western Australia prospective study in 1981-1998. The cohort consisted of the 1612 men and women aged 40-89 years who participated in the 1981 Busselton Health Survey and who were free of cardiovascular disease at that time. Serum ferritin levels were obtained from serum samples stored frozen since 1981. The outcomes of interest were time to first CHD event (hospital admission or death) and time to first stroke event. Case-cohort sampling was used to reduce costs and preserve serum but still allow efficient analysis. Ferritin assays were performed for 217 CHD cases, 118 stroke cases, and a random sample of 450 of the total cohort. Proportional hazards regression models were used to obtain age-adjusted and multivariate-adjusted hazard ratios for ferritin level in relation to CHD and stroke. The hazard ratio for the highest tertile group compared with the lowest group was 0.96 (95% confidence interval: 0.60, 1.53) for CHD and 1.43 (95% confidence interval: 0.78, 2.64) for stroke. Little or no evidence was found that ferritin level was a risk factor for cardiovascular disease.