Background New atrial pacing techniques and overdrive pacing algorithms have been introduced to prevent atrial fibrillation. This study was designed to test the hypotheses that (1) interatrial septum pacing (IASP) at the triangle of Koch would be more effective than right atrial appendage pacing (RAAP) in preventing paroxysmal atrial fibrillation (PAF) in patients with sinus bradycardia and (2) an algorithm (CAP) designed to achieve constant atrial capture would increase the efficacy of rate-responsive atrial pacing. Methods We studied 46 patients with PAF and sinus bradycardia implanted with a DDD(R) (Medtronic Thera) pacemaker. Twenty-four patients (6.0 ± 10.1 PAF episodes/month within 3 months before study) were randomized to RAAP and 22 patients (5.4 ± 7.1, not significant) to IASP. Within each arm 2 randomized crossover periods of CAP-OFF and CAP-ON function were programed. Results The PAF episodes per month significantly decreased in the RAAP (CAP-OFF: 2.1 ± 4.2, P < .05; CAP-ON: 1.9 ± 3.8, P < .05) and in the IASP group (CAP-OFF: 0.2 ± 0.5, P < .05; CAP-ON: 0.2 ± 0.5, P < .05). Values were significantly lower in the IASP group than in the RAAP group in both CAP-OFF (0.2 ± 0.5 vs 2.1 ± 4.2, P < .05) and CAP-ON (0.2 ± 0.5 vs 1.9 ± 3.8, P < .05) conditions. PAF burden was significantly lower in the IASP than in the RAAP group in CAP-OFF (47 ± 84 min/d vs 140 ± 217, P < .05) and in CAP-ON (41 ± 72 vs 193 ± 266, P < .05) conditions. No differences were observed within each arm in PAF burden between the 2 crossover CAP programing periods. Conclusions Rate-adaptive IASP at the triangle of Koch is more effective than RAAP in preventing PAF in patients with sinus bradycardia. In our sample of patients no additional clinical benefit is furnished by the CAP algorithm. (Am Heart J 2001;142:1047-55.)