The purpose of this study was to evaluate a pacing system using the recognition of cardiac evoked response for the automatic adjustment of pacing output. Patients were prospectively followed after primary implantation of VVIR pacemakers using AutoCapture (St. Jude Medical CRMD). Sensing and pacing thresholds, polarization signal, evoked response, and AutoCapture performance were evaluated with serial visits and 24-hour Holter monitoring. Three hundred ninety-eight patients (mean age 71 +/- 15 years) were followed for an average duration of 1 year (3 days-1.75 years) with the algorithm functional in > 90% of patients. Backup pacing in the event of exit block was confirmed in all patients. Pacing thresholds remained stable at 0.89 +/- 0.34 V with a pulse width of 0.31 ms (with chronic output autoset at 0.3 V above the actual threshold). Evoked response exhibited a small but statistically significant increase with time (8.92 mV at implant, 9.60 mV at 12 months), however, this finding did not result in any change in AutoCapture function during our follow-up period. The polarization signal remained stable with minimal variation (1.12 mV at implant, 1.18 at 12 months). No clinical adverse events were observed using the AutoCapture algorithm. In this initial experience with the AutoCapture algorithm the evoked response and polarization measurements remained adequate, allowing the system to function in the majority of patients with safe, low output pacing. High energy backup pacing provided an added safety feature over fixed output devices in cases of unexpected threshold rises. Longer follow-up is required for continued long-term validation of the algorithm.