To evaluate the incidence, prevalence, characteristics, and prognosis associated with clinically unrecognized myocardial infarction as diagnosed by electrocardiographic changes. Prospective, population-based cohort study with 4- to 20-year follow-up. Icelandic Heart Association Preventive Clinic. 9141 men residing in the Reykjavik area who were born between 1907 and 1934. Patients were assigned to categories of coronary heart disease at first visit on the basis of hospital records, Rose chest pain questionnaire, standardized 12-lead electrocardiogram, and history and physical examination. Incidence and prevalence of unrecognized myocardial infarction were determined, survival was measured, and causes of death were determined from death certificates and autopsy records. Prevalence was strongly influenced by age. Nearly undetectable in the youngest age group, it increased to more than 5% in the group aged 75 to 79 years. Incidence was almost zero up to age 40, then increased steeply to more than 300 cases per year per 100,000 persons at age 60, and decreased with age after age 65. Ten- and 15-year survival probabilities were 51% and 45%, respectively, and were similar to those for patients with recognized myocardial infarction. One third of men with unrecognized and 58% of men with recognized myocardial infarction had a history of angina pectoris (P < 0.001). Angina pectoris had a greater effect on coronary heart disease mortality in the former group than in the latter. The risk ratio for unrecognized myocardial infarction was 4.6 without angina (95% CI, 2.4 to 8.6) and 16.9 with angina (CI, 9.4 to 30.3); the risk ratio for recognized myocardial infarction was 6.3 without angina (CI, 3.7 to 10.6) and 8.5 with angina (CI, 5.8 to 12.6). At least one third of all myocardial infarctions were unrecognized. Prognosis and risk factor profiles for patients with recognized and unrecognized myocardial infarction were similar. Although those with unrecognized myocardial infarction were less likely than those with recognized myocardial infarction to have a history of angina pectoris, angina in these cases was usually associated with ischemic electrocardiographic changes and a poor prognosis, suggesting severe coronary heart disease.