R of sinus rhythm improves functional capacity and alleviates palpitations in patients with atrial fibrillation (AF). The advantages of direct-current cardioversion include the immediate restoration of sinus rhythm, as opposed to the unpredictable time to cardioversion in pharmacologic intervention, and the avoidance of potential adverse drug reactions. Although direct-current cardioversion is thought to have a higher success rate than pharmacologic intervention, it may fail to restore sinus rhythm in 5% to 30% of procedures. Few data are available regarding the association between clinical variables and the failure of direct-current cardioversion. Also, the relation between abnormalities of cardiac function and the success of direct-current cardioversion has not been studied in a large number of patients. This study assesses the success rate of direct-current cardioversion in patients with AF and the relation between clinical and echocardiographic variables and the success of direct-current cardioversion. • • • We studied 692 patients (459 men [66%] and 233 women [34%], mean age SD 67 13 years) with AF who had a first elective direct-current cardioversion at Mayo Clinic (Rochester, Minnesota). This study was approved by the Institutional Review Board of the Mayo Foundation. The duration of AF was 48 hours in 653 patients (94%) and 48 hours in 39 patients (6%). All patients underwent transthoracic echocardiography before cardioversion. Transesophageal echocardiography was performed in 201 patients (29%) before cardioversion. A history of coronary artery disease was considered present if the patient had had a previous myocardial infarction, myocardial revascularization, significant coronary artery stenosis on angiography, or typical anginal complaints. Valvular heart disease was considered present if the patient had more than mild stenosis or regurgitation of 1 cardiac valve. Idiopathic dilated cardiomyopathy was defined as left ventricular dilatation and global hypokinesia without significant coronary artery disease. The duration of AF was determined by the time of documentation of AF on the electrocardiogram or the onset of symptoms, whichever occurred earlier. Two-dimensional and M-mode imaging were performed with commercially available echocardiographic machines equipped with 2.5and 3.5-MHz phased-array transducers. M-mode echocardiography was used to measure cardiac dimensions and wall thickness. A left atrial diameter of 40 mm defined left atrial dilatation. This was considered mild for values 50 mm and marked for values 50 mm. Ejection fraction at rest was measured using a previously validated modification of the method of Quinones in Dujardin et al or by visual estimation. Structural heart disease was defined as any of the following: valvular heart disease, coronary heart disease, cardiomyopathy, or ejection fraction 50%. All procedures were performed with electrocardiographic monitoring and full equipment for cardiopulmonary resuscitation. Digitalis preparations were withheld for 24 to 48 hours before the procedure. After administration of intravenous sedation, synchronized direct-current monophasic shock was administered using the anteroposterior paddle position (sternal body angle of the left scapula). If sinus rhythm was not restored after the first shock, the procedure was repeated (up to 4 times) using a larger number of joules (up to 360 J). The number of joules with which cardioversion was initiated was determined for each patient according to physician discretion. Electrocardiography was performed before and after the procedure to verify cardiac rhythm. Successful cardioversion was defined as restoration of sinus rhythm after administration of direct-current shock. Unless specified, data are presented as mean values SD or as frequency percentages. The rank sum and chi-square tests were used to compare differences between continuous variables and proportions, respectively. Univariate and multivariate stepwise logistic regression models were used to identify individual and joint predictors of unsuccessful cardioversion. Analyses of success were done using both the patient and the individual shock as the unit of observation to harvest information when multiple shocks were needed. Differences were considered significant if the null hypothesis could be rejected at the 0.05 probability level. These diseases were present in the study population: systemic hypertension in 355 patients (51%), coronary artery disease in 233 patients (34%), valvular heart disease in 227 patients (33%), and cardiomyopathy (idiopathic dilated, restrictive, or hypertrophic) in 125 patients (18%). Cardioversion was successful in 592 patients (86%) and unsuccessful in 100 (14%). The mean number of shocks was 2.1 1.2/patient. The mean number of joules was 222 100. In 74 patients, 50 J was used initially, with success achieved From the Division of Cardiovascular Diseases and Internal Medicine and the Section of Biostatistics, Mayo Clinic, Rochester, Minnesota. Dr. Khandheria’s address is: Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905. Manuscript received June 13, 2001; revised manuscript received and accepted September 6, 2001.