BACKGROUND Atrioventricular (AV) node ablation and pacing has become accepted therapy for drug-refractory paroxysmal atrial fibrillation (PAF). However, few data demonstrate its superiority over continued medical therapy. The influence of pacing mode and mode-switch algorithm has not been investigated. METHODS AND RESULTS Symptomatic patients who had tried >/=2 drugs for PAF were randomized to continue medical therapy (n=19) or AV junction ablation and implantation of dual-chamber mode-switching (DDDR/MS) pacemakers (slow algorithm [n=19] or fast algorithm [n=18]). Follow-up over 18 weeks was at 6-week intervals and used quality-of-life questionnaires (Psychological General Well Being [PGWB], McMaster Health Index [MHI], cardiac symptom score), exercise testing, echocardiography, and Holter monitoring. Paced patients were randomized to DDDR/MS or VVIR and subsequently crossed over. Ablation and DDDR/MS pacing produced better scores than drug therapy for overall symptoms (-41%, P<0.01), palpitations (-58%, P=0. 0001), and dyspnea (-37%, P<0.05). Changes in score from baseline were better with ablation and DDDR/MS pacing for overall symptoms (-48% versus -4%, P<0.005), palpitation (-62% versus -5%, P<0.001), dyspnea (-44% versus -3%, P<0.005), and PGWB (+12% versus +0.5%, P<0. 05). DDDR/MS was better than VVIR pacing for overall symptoms (-21%, P<0.05), dyspnea (-30%, P<0.005), and MHI (+5%, P<0.03). There were no differences between algorithms. More patients developed persistent AF with ablation and pacing than with drugs at 6 weeks (12 of 37 versus 0 of 19, P<0.01). CONCLUSIONS Ablation and DDDR/MS pacing produces more symptomatic benefit than medical therapy or ablation and VVIR pacing but may result in early development of persistent AF.