Background— Noncontact mapping (NCM) has not been validated as a clinical technique to measure ventricular repolarization. We used NCM to determine repolarization characteristics by analysis of reconstructed unipolar electrograms (UEs) at the same sites as monophasic action potential (MAP) recordings in the human ventricle. Methods and Results— MAPs were recorded from a total of 355 beats at 46 sites in the left or right ventricle of 9 patients undergoing ablation of ventricular tachycardia guided by NCM (EnSite system). Measurements were made during sinus rhythm, constant right ventricular pacing, and ventricular extrastimuli during restitution-curve construction. The EnGuide locator signal was used to document MAP catheter locations on the endocardial geometry. UE-determined activation-recovery interval (ARI) measured at the maximum derivative of the T wave (Wyatt method) and the minimum derivative of the positive T wave (alternative method) was correlated with MAP measured at 90% repolarization (MAP90%) at the same sites. ARI correlated with MAP90% during steady state by the Wyatt method ( r =0.83, P <0.001) and the alternative method ( r =0.94, P <0.001). Restitution curves constructed from MAP and UE data exhibited the same characteristics, with a mean correlation coefficient of 0.95 (range, 0.90 to 0.99, P <0.001). The error between ARI and MAP90% was greater over a shorter diastolic coupling interval but was not influenced by distance of the sampling site from the multielectrode array. Conclusions— NCM accurately determines steady-state and dynamic endocardial repolarization in humans. Global, high-density, NCM data could be used to characterize abnormalities of human ventricular repolarization.