To assess immediate and long term success of "fast" pathway catheter ablation with graded use of radiofrequency energy in patients with classic atrioventricular nodal reentrant tachycardia (AVNRT) and evaluate clinical, procedure related, and electrophysiological features affecting long term results. 31 consecutive patients with classic AVNRT at electrophysiological study, who were candidates for radiofrequency ablation. Patients were followed for an average of 24 months after ablation. All studies and ablations were performed in an electrophysiological laboratory under fluoroscopic guidance using standard electrophysiological techniques. Radiofrequency application was performed at the site of proximal His bundle electrogram with A:V ratio of > 1. It was started at 10 W with increment of 5 W to a maximum of 25 W at 60 s. With the onset of junctional rhythm, atrial pacing was begun in order to monitor the PR interval. Application was terminated prematurely with a non-conducted P wave, continued prolongation of the PR interval beyond 50% of the baseline, or a threefold rise in impedance. Successful ablation was possible in 30/31 patients (97%) with an average of seven applications (range 1-10). It was associated with significant prolongation of PR interval (P < 0.001) and AV Wenckebach cycle length (P = 0.01). Ventriculo-atrial conduction was abolished in 24/30 patients (82%) with successful ablation. Two patients developed transient complete heart block (3 and 12 min) and one persistent right branch block. Four patients had late recurrence. Presence of ventriculo-atrial block was the only electrophysiological index predictive of long term success (P = 0.01). Graded use of radiofrequency energy and atrial pacing to monitor PR interval decreases the risk of atrioventricular block in patients undergoing fast pathway ablation for AVNRT. Ventriculo-atrial block is predictive of long term success and should be a preferred end point for fast pathway ablation.