A case-control study was conducted to investigate the association between serum selenium and risk of death from acute coronary heart disease (CHD) as well as risk of fetal and non-fetal myocardial infarction (MI). Case-control pairs came from a population of 11,000 persons examined in 1972 from two counties in eastern Finland, an area with an exceptionally high mortality from cardiovascular diseases. Cases were aged 35-59 years and had died of CHD or other CVD or had a non-fetal MI during a seven-year follow-up. Controls were matched for sex, age, daily tobacco consumption, serum cholesterol, diastolic blood pressure, and history of angina pectoris. The mean serum selenium concentration for all cases was 51.8 micrograms/l and for all controls 55.3 micrograms/l (p less than 0.01). Serum selenium of less than 45 micrograms/l was associated with an adjusted relative risk of CHD death of 2.9 (p less than 0.01, 95% CI, 1.4-6.0), a relative risk of CVD death of 2.2 (p less than 0.01, 95% CI, 1.2-4.0), and a relative risk of fatal and nonfatal MI of 2.1 (p less than 0.001, 95% Ci, 1.4-3.1). 22% (95% CI, 8-35%) of contrary deaths were attributable to serum selenium in the whole study population.