Whether an association, causative or not, exists between the level of serum selenium and the risk of ischaemic heart disease (IHD) remains unsettled. We investigated the issue in a cohort of 3387 males aged 53-74 years (mean 63). Based on information about health status, life-style and socioeconomic factors given in a prefilled comprehensive questionnaire, the men were interviewed and the information validated. Following the interview, they underwent a clinical examination and had a venous blood sample drawn for the determination of a number of biochemical characteristics. Three hundred and forty-six men were excluded due to prevalent cardiovascular disease, including stroke. During the next three years (1986-1989) 107 men (approximately 3%) suffered an IHD event; 25 events were fatal. Compared to others, men with serum selenium levels less than or equal to 1 mumol/l, approximately the lowest tertile, had a 70% increased risk of IHD, relative risk (RR) with 95% confidence limits was 1.70 (1.14-2.53). After multivariate adjustment for cholesterol, social class, smoking and age, RR was 1.55 (1.00-2.39). Serum selenium level was significantly (P less than 0.05), but not strongly, correlated with a number of IHD risk factors: serum cotinine, tobacco smoking, social class, alcohol consumption, total cholesterol, hypertension, age and physical inactivity. Body mass index, HDL-cholesterol and triglycerides were not significantly associated with serum selenium. We conclude that middle-aged and elderly Danish men with serum selenium less than or equal to 1 mumol/l had a significantly increased risk of ischaemic heart disease. This association was not explained by the interrelationship of serum selenium and major cardiovascular risk factors.