The prevalence, incidence and prognosis of recognized and unrecognized Q-wave myocardial infarction (MI) was assessed in an 8-year prospective study of 390 community-based subjects (age 75 to 85 years at entry, mean 79 years). Subjects were studied at baseline and with annual follow-up electrocardiographic (ECG) exams. At baseline, 7.9% had a history of MI without ECG evidence, 6.4% had ECG evidence of Q-wave MI without clinical history, 4.1% had both clinical history and ECG evidence and 81.5% had neither history nor ECG evidence (control subjects). After an average follow-up period of 76.2 months, the total mortality rate was 5.9/100 person-years for subjects with some evidence of MI at baseline versus 3.9 in the control group (p = 0.059). The incidence of cardiovascular disease in subjects with evidence of MI was 8.8/100 person-years versus 4.7 among control subjects (p = 0.002). During the follow-up period, 115 new Q-wave MIs occurred (50 unrecognized, rate 2.4/100; 65 recognized, rate 3.2/100). There was no difference in mortality and morbidity outcome between subjects with recognized and unrecognized MIs. Those with only a history of MI at baseline had a threefold greater risk of a new MI (recognized and unrecognized) than the control group (p = 0.003). Unrecognized Q-wave MI is a common occurrence in the "old old" with subsequent morbidity and mortality prognosis comparable to that of recognized MI. History of MI alone in this age group is also associated with an increased risk of MI, suggesting the need for better diagnostic markers of myocardial ischemia in the old.