The effects of antiarrhythmic drugs on electrograms have implications for arrhythmia-detection algorithms in implantable antitachycardia devices. Filtered and unfiltered intra-atrial electrograms were analyzed in eight patients who received procainamide (50 mg/min iv, up to 1000 mg) during 11 episodes of atrial fibrillation. Continuous recordings were made before, during, and after the infusion. The recordings were digitized, divided into 4.27 sec segments, and analyzed for atrial rate, median frequency and amplitude probability density function. Significant differences were noted before and after infusion of procainamide for atrial rate (498 +/- 97 vs 356 +/- 146 beats/min; p less than .005), median frequency (5.50 +/- 1.22 vs 4.24 +/- 0.99 Hz; p less than .0005), and density (58.3 +/- 13.9% vs 69.1 +/- 15.0%; p less than .005). Pre- and postprocainamide values were compared with published criteria for detection of atrial fibrillation. Before procainamide, only 2.3%, 5.7%, and 3.4% of the data segments failed to meet criteria for atrial fibrillation by rate, frequency content, and density, respectively. In contrast, after procainamide, 50%, 36.4%, and 28.4% of the data segments failed to meet these same criteria, despite electrograms still meeting morphologic criteria for atrial fibrillation. Thus procainamide resulted in changes sufficient to cause failure of published criteria for detection of atrial fibrillation. These findings have broad implications for the function of antitachycardia devices in patients receiving antiarrhythmic drug therapy.