Atrial fibrillation is a common arrhythmia associated with increased risk for embolic stroke. Restoration of sinus rhythm in patients with atrial fibrillation is a logical strategy to prevent the cardiovascular and thromboembolic complications of this dysrhythmia. The most common strategy for restoration of sinus rhythm is pharmacological antiarrhythmic therapy with or without electrical cardioversion. Five randomized clinical trials compared rhythm to rate-control strategies in patients with atrial fibrillation. These trials examined mortality, thromboembolic complications, exercise tolerance, quality of life, hospital admissions and drug-related adverse reactions. Mortality ranged from 2.9% to 23.8% among the trial subjects randomized to rhythm control versus 1.0% to 21.3% in the rate control subjects. The risk of thromboemboli was greater: 2.9% to 7.9% in the rhythm-control subjects compared with 0% to 5.5% in the rate control subjects. Hospital admissions and drug-related adverse events were increased in the rhythm-control subjects. Stroke and systemic emboli occurred more often in the rhythm-control subjects many of whom had been withdrawn from anticoagulation. Rhythm-control offered no advantage compared with rate control for patients with atrial fibrillation at increased risk for stroke. One explanation for this finding is that those patients thought to have been successfully converted to sinus rhythm in fact had asymptomatic paroxysmal episodes of atrial fibrillation increasing their risk of stroke because they were unprotected by anticoagulation. Pharmacological attempts to restore atrial fibrillation to sinus rhythm do not improve mortality or reduce thromboembolic events. All patients with atrial fibrillation at increased risk for stroke should be continued on long-term anticoagulation even if they appear to have been successfully restored to sinus rhythm.