We studied the effect of diabetes mellitus (DM) on mortality rate in 42,595 patients in the statewide Myocardial Infarction Data Acquisition System (MIDAS), which included patients with myocardial infarction from 90 nonfederal hospitals in New Jersey during the years 1986 and 1987. Of these patients 9695 (22.8%) had DM. DM was more prevalent among female, black, and older patients. DM was associated with higher mortality rates, both in-hospital (21.5% vs 19.2%, p < 0.001) and during 3-year follow-up (46.7% vs 37.8%, p < 0.001). This relation persisted in both men and women, blacks and whites, and all age groups. DM was an independent predictor of mortality by multivariate Cox proportional hazards regression analysis after adjustments were made for gender, race, age, hypertension, left ventricular dysfunction, chronic pulmonary disease, chronic liver disease, and anemia. This effect of DM was most pronounced in the younger age groups. Relative risk was 1.87 for age group 30 to 49, 1.36 for 50 to 69, and 1.17 for 70 to 89 years (p < 0.0001).