To determine the evolution of acute myocardial infarction in patients with diabetes we study 207 consecutive patients with myocardial infarction. Using WHO's criteria 23% of our cases were diagnosed of diabetes mellitus. Diabetic patients were older than non diabetic (67.9 +/- 10 years vs. 62.4 +/- 11 years, p less than 0.05) and had a higher ratio of females (52% vs. 21%, p less than 0.001). Cigarette smoking was infrequent in diabetic population. Incidence of other risk factors was comparable. Despite an increased proportion on no q-wave myocardial infarction in the diabetic patients (12.5% vs. 6.9%, p NS), the site of infarction was similar into the two groups. Acute phase mortality was higher in the diabetic group (37.5% vs. 16.3%, p less than 0.001). This increased mortality is, partially, related to an increased incidence of pump failure, but a multivariate analysis using stepwise logistic regression, selected diabetes as an independent predictor of prognosis. Survivors were followed for 41 +/- 20 months; diabetic patients showed a poor prognosis with a higher incidence of congestive heart failure (42.8% vs. 13.7%, p less than 0.01), reinfarction (16.6% vs. 8.5%) and death. Cox proportional hazard model selected diabetes as an independent predictor of survival. We conclude that patients with diabetes mellitus constitute a subgroup into the myocardial infarction population; this subgroup had greater mortality than non diabetic patients in relation to increased incidence of pump failure, but multivariate analysis indicates that other factors not considered in the present study may play a role in their poor prognosis.