Direct current electric shocks have been used to terminate atrial arrhythmias (cardioversion) in humans since the 1960s. The likelihood of successful cardioversion and maintenance of sinus rhythm is increased if the left atrium is not markedly enlarged and fibrotic, if there is no marked left atrial hypertension (e.g., mitral stenosis), and if the arrhythmia is not long-standing. To minimize the risk of thromboembolic phenomena, therapeutic anticoagulation should be established for at least 3 weeks before and for 4 weeks after cardioversion; coumadin is usually used for this purpose. A more recent approach uses transesophageal echocardiography to demonstrate the absence of thrombi in the left atrium and left atrial appendage. If no thrombi are evident, 48 hours of heparin anticoagulation may be adequate prior to cardioversion. Anticoagulation is still required after cardioversion. Quinidine and digitalis, singly or in combination, are frequently used to achieve and maintain sinus rhythm in association with cardioversion. For the procedure itself, traditional hand-held paddle electrodes or self-adhesive electrode pads may be used; the apex-anterior and anterior-posterior positions are equally effective. Gel couplants and firm pressure should always be used with hand-held paddles to reduce transthoracic impedance and maximize current flow. Electrodes should be widely separated to avoid shunting of current along the chest wall between electrodes. Generally, electrodes should be large in size; small "pediatric" electrodes should only be used in infants < 1 year of age (< 10 kg). Shocks should always be synchronized to the R wave to avoid the vulnerable period and the inadvertent induction of ventricular fibrillation. Initial shocks for atrial fibrillation should begin at 100 J; atrial flutter generally requires a smaller shock (initial shocks at 50 J). Effective anesthesia, not merely sedation, is required to achieve amnesia and avoid pain. Exciting new developments in defibrillation and cardioversion have occurred. It is now understood that excessive energy and current may induce cardiac damage, and recent studies suggest such damage may be mediated in part by free radicals. New shock waveforms, such as biphasic and multiphasic waveforms from multiple encircling electrodes, may be superior to the standard damped sinusoidal waveform.