BACKGROUND: Atrial fibrillation increases stroke risk and adversely affects cardiovascular haemodynamics. Electrical cardioversion may, by restoring sinus rhythm, improve cardiovascular haemodynamics, reduce the risk of stroke, and obviate the need for long-term anticoagulation. OBJECTIVES: To assess the effects of electrical cardioversion of atrial fibrillation or flutter on the risk of thromboembolic events, strokes and mortality (primary outcomes), the rate of cognitive decline, quality of life, the use of anticoagulants and the risk of re-hospitalisation (secondary outcomes) in adults (>18 years). SEARCH STRATEGY: We searched the Cochrane CENTRAL Register of Controlled Trials (1967 to May 2004), MEDLINE (1966 to May 2004), Embase (1980 to May 2004), CINAHL (1982 to May 2004), proceedings of the American College of Cardiology (published in Journal of the American College of Cardiology 1983 to 2003), www.trialscentral.org, www.controlled-trials.com and reference lists of articles. We hand-searched the indexes of the Proceedings of the British Cardiac Society published in British Heart Journal (1980 to 1995) and in Heart (1995 to 2002); proceedings of the European Congress of Cardiology and meetings of the Joint Working Groups of the European Society of Cardiology (published in European Heart Journal 1983-2003); scientific sessions of the American Heart Association (published in Circulation 1990-2003). Personal contact was made with experts. SELECTION CRITERIA: Randomised controlled trial or controlled clinical trials of electrical cardioversion plus 'usual care' versus 'usual care' only, where 'usual care' included any combination of anticoagulants, antiplatelet drugs and drugs for 'rate control'. We excluded trials which used pharmacological cardioversion as the first intervention, and trials of new onset atrial fibrillation after cardiac surgery. There were no language restrictions. DATA COLLECTION AND ANALYSIS: For dichotomous data, odds ratios were calculated; and for continuous data, the weighted mean difference was calculated. MAIN RESULTS: We found three completed trials of electrical cardioversion (rhythm control) versus rate control, recruiting a total of 927 participants (Hot Cafe; RACE; STAF) and one ongoing trial (J-RHYTHM). There was no difference in mortality between the two strategies (OR 0.83; CI 0.48 to 1.43). There was a trend towards more strokes in the rhythm control group (OR 1.9; 95% CI 0.99 to 3.64). At follow up, three domains of quality of life (physical functioning, physical role function and vitality) were significantly better in the rhythm control group (RACE 2002; STAF 2003). AUTHORS' CONCLUSIONS: Electrical cardioversion (rhythm control) led to a non-significant increase in stroke risk but improved three domains of quality of life.