Atrial pacing has been shown to delay the onset of atrial fibrillation (AF) when compared with ventricular pacing in patients with sick sinus syndrome. The role for pacing in the control of AF in patients without bradycardia is uncertain. We performed a randomized, crossover, single-blinded study in 22 patients (14 women, aged 63 +/- 10 years) with paroxysmal AF refractory to treatment with oral sotalol (202 +/- 68 mg/day) and no bradycardic indication for pacing. All patients received a dual-chamber pacemaker with 2 atrial pacing leads positioned at the high right atrium and coronary sinus ostium, respectively. Patients were randomized in a crossover fashion to be paced for 12 weeks, either with high right atrial (RA) pacing at 30 beats/min ("Off") or dual-site RA pacing with an overdrive algorithm that maintained atrial pacing at a rate slightly above the sinus rate ("On"). Treatment on resulted in a significantly higher percentage of atrial pacing and a reduction in atrial ectopic frequency than the treatment off period. The time to the first clinical AF recurrence was prolonged (15 +/- 17 to 50 +/- 35 days, p = 0.006), and total AF burden was reduced (45 +/- 34% vs 22 +/- 29%, p = 0.04) in the on-treatment phase. However, there was no difference in AF checklist symptom scores or overall quality-of-life measures. Dual-site RA pacing with continued sinus overdrive prolonged the time to AF recurrence and decreased AF burden in patients with paroxysmal AF. The absence of a major impact on symptom control suggests that pacing should be used as an adjunctive therapy with other treatment modalities for AF.