To assess the effect of diabetes on outcome after acute myocardial infarction (MI), we compared a cohort of 228 type II (non-insulin-dependent) diabetic patients who had sustained acute MI with a similar number of nondiabetic patients with MI. Thirty-day mortality was greater in the diabetic group (27 vs. 17%). However, diabetic patients were older and had more cardiovascular disease before MI. Analyses accounting for such baseline risk revealed a complex effect of diabetes. The relative risk (RR) of dying from MI due to diabetes was greatest among patients with lowest baseline risk (RR 7.3) and least among those at highest baseline risk (RR 0.83). These effects were most striking with transmural MI, which was highly lethal for those with diabetes. Analyses with pulmonary edema as the endpoint support the significant risk conferred by diabetes and its interaction with baseline risk. Diabetes is a risk factor for poor outcome after MI, particularly among patients whose pre-MI cardiovascular status otherwise appears normal.