BACKGROUND: To compare in atrial fibrillation patients with and without diabetes, (1) baseline characteristics, (2) additional risk factors for stroke or peripheric or visceral embolism (hypertension, previous stroke, age >75 years), (3) mortality, (4) stroke or embolism, and (5) oral anticoagulation in the year 2000. METHODS: Included were 409 outpatients with nonrheumatic atrial fibrillation (62 +/- 12 years, 36% female). All underwent transthoracic and transesophageal echocardiography. Patients with thrombi received oral anticoagulation; patients without thrombi received aspirin until the follow-up in 1995; afterwards, oral anticoagulation according to risk factors for stroke or embolism was recommended. Patients were contacted during the year 2000. RESULTS: Type 2 diabetes was diagnosed in 73 patients (18%). Sixteen (22%) diabetic and 169 (50%) nondiabetic patients had no other risk factors for stroke or embolism (p < 0.0001). Diabetic patients were older, had more frequent heart failure, hypertension, myocardial infarction, left ventricular dysfunction, valvular abnormalities, left atrial or appendage thrombi, larger left atria, and left atrial appendages than nondiabetic patients. Mean follow-up was 115 months. Diabetic patients had a higher mortality than nondiabetic patients (7%/year versus 4%/year, p < 0.0001). The rate of stroke or embolism of diabetic (3%/year) and nondiabetic patients (2%/year) was similar. The rate of oral anticoagulation was higher in diabetic than in nondiabetic patients (p = 0.0066). CONCLUSIONS: Diabetic patients with atrial fibrillation frequently have additional risk factors for stroke or embolism, and thus should be treated with oral anticoagulation. Whether in the rare cases of atrial fibrillation, in whom diabetes is the only clinical risk factor, oral anticoagulation is indicated cannot be answered by the present study.