Thirteen patients in chronic atrial fibrillation with a normal resting heart rate but with exercise tachycardia and episodes of bradycardia were randomised to treatment periods of two weeks on xamoterol (200 mg twice daily), low dose digoxin, or placebo, in a blind crossover study. The results (mean SEM) of symptom scores, a treadmill exercise test, and 24 hour ambulatory electrocardiographic monitoring were obtained. Xamoterol improved symptom scores and controlled exercise heart rate better than digoxin. Xamoterol was better than digoxin or placebo in reducing the heart rate response to exercise and tended to improve exercise duration. Xamoterol, by reducing the daytime maximum hourly heart rate and increasing the night time minimum hourly heart rate, significantly reduced the difference between the two compared with placebo. In contrast, digoxin tended to reduce both the maximum and minimum hourly heart rates through day and night. Both the frequency and duration of ventricular pauses were reduced by xamoterol but tended to increase with digoxin. Xamoterol reduced both the circadian variation in ventricular response to atrial fibrillation and exercise tachycardia by modulating the heart rate according to the prevailing level of sympathetic activity. These changes were translated into symptomatic benefit for the patients studied.