The relation between plasma insulin level and the prevalence of coronary heart disease was studied in a representative nondiabetic population sample of 396 men and 673 women aged 65-74 years in Kuopio, eastern Finland, in 1986-1988. No association between plasma insulin levels and the prevalence of previous myocardial infarction was seen, except for a positive relation between 2-hour insulin level and myocardial infarction in women (4.1, 4.1, and 11.9%, p < 0.01; insulin quintiles I+II, III+IV, and V, respectively). However, the prevalence of ischemic electrocardiographic changes increased by quintiles of fasting plasma insulin (not significant) and 2-hour plasma insulin (39.8, 50.6, and 59.5% in men, p < 0.01; 35.9, 39.0, and 53.0% in women, p < 0.01). Furthermore, the prevalence of angina pectoris increased by fasting insulin (19.3, 19.0, and 33.8%, in men, p < 0.05; not significant in women) and 2-hour insulin quintiles (11.2, 28.8, and 30.4% in men, p < 0.001; 17.0, 17.8, and 28.4% in women, p < 0.05). There was a marked clustering of risk factors, including high total triglyceride levels, low high density lipoprotein cholesterol levels, and hypertension in subjects in the highest insulin quintiles. The association between insulin levels and coronary heart disease, except for myocardial infarction in women, diminished in multivariate analyses. In conclusion, hyperinsulinemia was related to an increased prevalence of coronary heart disease in elderly subjects which may be due to in part to adverse alterations in other risk factors associated with hyperinsulinemia.