OBJECTIVE:This study sought to determine whether stress echocardiography using exercise (when feasible) or dobutamine echo could be used to predict mortality in patients with diabetes. RESEARCH DESIGN AND METHODS:Stress echo was performed in 937 patients with diabetes (aged 59 +/- 13 years, 529 men) for symptom evaluation (42%) and follow-up of known coronary artery disease (CAD) (58%). Stress echocardiography using exercise was performed in 333 patients able to exercise maximally, and dobutamine echo using a standard dobutamine stress was used in 604 patients. Patients were followed for < or = 9 years (mean 3.9 +/- 2.3) for all-cause mortality. RESULTS:Normal studies were obtained in 567 (60%) patients; 29% had resting left ventricular (LV) dysfunction, and 25% had ischemia. Abnormalities were confined to one territory in 183 (20%) patients and to multiple territories in 187 (20%) patients. Death (in 275 [29%] patients) was predicted by referral for pharmacologic stress (hazard ratio [HR] 3.94, P < 0.0001), ischemia (1.77, P < 0.0001), age (1.02, P = 0.002), and heart failure (1.54, P = 0.01). The risk of death in patients with a normal scan was 4% per year, and this was associated with age and selection for pharmacologic stress testing. In stepwise models replicating the sequence of clinical evaluation, the predictive power of independent clinical predictors (age, selection for pharmacologic stress, previous infarction, and heart failure; model chi(2) = 104.8) was significantly enhanced by addition of stress echo data (model chi(2) = 122.9). CONCLUSIONS:The results of stress echo are independent predictors of death in diabetic patients with known or suspected CAD. Ischemia adds risk that is incremental to clinical risks and LV dysfunction.