We prospectively analyzed detection (D) and redetection (R) characteristics between transvenous defibrillators (ICD) using standard bipolar and integrated bipolar sensing. Monophasic and biphasic ICDs were included. Subthreshold shocks were intentionally delivered and redetection assessed at the predischarge, 1, 3, and 6 months follow-up retesting. Among the 160 ICDs with standard bipolar sensing 530 VF inductions were analyzed. After the failed shocks undersensing was more frequent (3% vs 20%, p < 0.01) but did not remarkably prolong redetection (3.13 ± 0.8 sec vs 3.3 ± 1.1 sec). Among the 150 ICDs with integrated bipolar sensing 60 were CPI and 90 Ventritex. After 200 failed shocks redetection prolonged with the CPI system (3.1 ± 1.4 sec vs 4.6 ± 3.6 sec, p < 0.05) while did not change after 270 failed shocks with the Ventritex ICD (5.3 ± 2.8 sec vs 4.9 ± 1.7 sec). This may reflect different nominal settings for 0 and R. In 6 out of 90 patients with Ventritex and lout of 60 with the CPI ICDs the device failed to redetect VF. None of those patients had sudden death at follow-up. Analysis of endocardial electrograms showed that failure to redetect VF is not associated with a uniform reduction, but with a rapid and repetitive change of electrogram amplitude. Conclusion 1) Standard bipolar sensing more effectively redetects VF; 2) endocardial electrogram analysis provide insights in understanding the mechanism of undersensing; 3) the clinical relevance of this phenomenon remains unknown.