To evaluate the response of patients with chronic atrial fibrillation to exercise, 50 men (mean age 65 ± 8 years) with atrial fibrillation underwent a maximal exercise test using respiratory gas exchange techniques. Patients were classified by the presence (n = 29) or absence ("lone atrial fibrillation," (n = 21) of underlying heart disease. Responses were evaluated at a standard submaximal work load (3.0 mph. [4.8 km/h] 0% grade), at the gas exchange anaerobic threshold and at maximal exercise. For all 50 patients, the mean maximal oxygen uptake was 20.6 ml/kg per min, which approximates 85% of the aerobic capacity predicted for age-matched normal individuals. Patients with lone atrial fibrillation demonstrated normal exercise capacity in contrast to patients with atrial fibrillation and known heart disease (22.7 ± 5 versus 19.1 ± 5.0 ml/kg per min, p < 0.05). The mean maximal heart rate (176 ± 30 beats/min) was approximately 20 beats/min higher than that expected for age, was extremely variable and accounted for only 8% of the variance in maximal oxygen uptake. Maximal heart rate in subjects with lone atrial fibrillation was higher than that of subjects with atrial fibrillation and known heart disease (189 ± 32 versus 166 ± 24 beats/min, p < 0.01). Stepwise regression analysis revealed that maximal systolic blood pressure accounted for 19% of the variance in maximal oxygen uptake (VO2 max), suggesting that systolic function is an important determinant of exercise performance in atrial fibrillation. It is concluded that 1) the exercise response in patients with lone atrial fibrillation differs markedly from the typical heterogeneous group of patients with atrial fibrillation and underlying heart disease, 2) the higher maximal heart rate observed in patients with lone atrial fibrillation may be a compensation for the loss of atrial function, and 3) exercise impairment in patients with atrial fibrillation is due to underlying heart disease and not the arrhythmia itself.