The accuracy of information retrievable from the memories of DDDR pacing systems has been limited by the absence of actual electrograms confirming the proper sensing of spontaneous cardiac activity versus that of extraneous signals. This study examined the diagnostic power of a new arrhythmia interpretation scheme, which includes the recording and storage of high resolution endocavitary atrial and ventricular electrograms (HREGM). HREGM stored in the memories of new generation pacemakers (PM) in response to nonsustained ventricular tachycardia (NSVT), sustained VT, and atrial arrhythmias were analyzed in a follow-up registry of 520 patients at 1 month, and 3 to 6 months after implantation of a PM for standard indications. For each sequence of stored HREGM, the accuracy of the PM response was examined, classified as accurate (true positive), versus inaccurate (false positive), versus undetermined, and the relative contribution of the HREGM in verifying the PM diagnosis was measured. During a follow-up of 4.9 +/- 2 months, 256 (49%) of the 520 patients had an event recorded, which was confirmed to be arrhythmic on the basis of HREGM. Overall, approximately 34% of atrialtachy response (ATR) episodes were confirmed to be appropriate. Similar percentages of episodes were prompted by oversensing of signals unrelated to cardiac arrhythmias, while nearly 12% of the episodes could not be clarified because of such brief duration as to preclude recording of their onset. Approximately one-third of NSVT, and one-half of VT detections were false positive. Ventricular oversensing, most often due to myopotential interference in presence of unipolar sensing, and atrial undersensing were both identified as sources of false-positive detections of ventricular events. The proportion of true-positive detections was significantly higher in the bipolar (83%) than unipolar configuration. Among 520 PM recipients, miscellaneous episodes of atrial arrhythmias were confirmed by HREGM in 37% of patients within 3 to 6 months of follow-up. Atrial fibrillation was confirmed in only 6% of patients, of whom over 50% already had histories of atrial fibrillation. The prevalence of unsuspected atrial arrhythmia in this unselected population was lower than previously reported.