Background Therapy for chronic atrial fibrillation (AF) focuses on rate versus rhythm control, but little is known about the effects of common therapeutic interventions on exercise tolerance in AF. Methods Six hundred fifty-five patients with chronic AF underwent maximal exercise testing at baseline and 8 weeks, 6 months, and 1 year after randomization to sotalol, amiodarone, or placebo therapy and attempted direct current cardioversion. Analyses of baseline determinants of exercise capacity, predictors of change in exercise capacity at 6 months and 1 year, and the short- and long-term effects of cardioversion on exercise capacity were made. Results Age, obesity, and presence of symptoms accompanying AF were inversely associated with baseline exercise capacity, but these factors accounted for only 10% of the variance in exercise capacity. Patients most likely to benefit from cardioversion were those most limited initially, younger, not obese or hypertensive, and with an uncontrolled ventricular rate at baseline. Conversion to sinus rhythm (SR) resulted in significant reductions in resting (≈25 beat/min) and peak exercise (≈40 beat/min) heart rates at 6 months and 1 year ( P < .001). Successful cardioversion improved exercise capacity by 15% at 8 weeks, and these improvements were maintained throughout the year. This improvement was observed both among those who maintained SR and those with intermittent AF. Conclusion Cardioversion resulted in a sustained improvement in exercise capacity over the course of 1 year, and this improvement was similar between those in SR and those with SR and recurrent AF. Patients most likely to improve with treatment tended to be younger and nonobese and have the greatest limitations initially.