We compare the clinical features and hospital outcomes in 83 diabetic patients admitted with acute myocardial infarction and 380 nondiabetic patients with levels of glycosylated hemoglobin (HbA1c) low enough to exclude undiagnosed diabetes. The hospital mortality was 42.2% in diabetic and 24.7% in nondiabetic patients, an odds ratio of 2.22 (CI 1.37-3.60, P less than .002). The excess mortality was due to cardiogenic shock and left ventricular failure (pump failure). There was no difference in peak levels of aspartate transaminase between the groups. Among the diabetic patients, the admission levels of plasma glucose and peak levels of aspartate transaminase were higher among those who developed pump failure or died, but there was no relationship between outcome and gender, disease duration, or treatment. Prior blood glucose control, as judged by levels of HbA1c, was not related to hospital outcome (P greater than .5). In a further study, the 83 diabetic patients were compared with 249 age- and sex-matched diabetic subjects without myocardial infarction for treatment, disease duration, and control. There was an increased risk of admission with myocardial infarction of 2.35 (CI 1.41-3.92, P less than .005) within the first 5 yr of diagnosis of diabetes. Infarct patients had significantly lower levels of HbA1c than control subjects (P less than .005), but treatment did not differ between groups. Neither incidence nor case fatality of myocardial infarction in diabetic patients is positively associated with cumulative glycemic exposure.