Background. Although ECG imaging technology has been in development for many years, the clinical validation of epi-endocardial ECGI mapping has been performed on relatively small groups of patients. Aim. To validate epi-endocardial ECGI mapping accuracy for different pacings with implanted pacemakers in single center single-blind cross-sectional study Methods. 30 patients (median age 61 years; min-max 27-78 years, 26 male) with previously implanted pacemakers underwent epi-endocardial ECGI mapping using “Amycard 01C EP Lab” system (EP Solutions SA, Switzerland). 240-channell ECG were recorded during right endocardial and left epicardial ventricular pacings followed by torso and ECG-gated cardiac computed tomography. The data obtained from CT was imported into “Amycard 01C EP lab” software in DICOM-format to reconstruct 3D polygonal models of the torso and heart. In this study we consider both epicardial and epi-endocardial 3D models of the heart. Afterwards epi- and endocardial unipolar electrograms as well as isopotential maps were reconstructed using an iterative Tikhonov regularization algorithm. Early activation zone was determined from the isopotential maps by 3 independed investigators (2 blinded and 1 non-blinded with experience from 6 months to 5 years in ECGI). A p-value < 0.01 was considered as statistically significant. Statistical analysis was performed using Statistica v.12 (Statsoft Inc, USA). Results. The mean (SD) geodesic distance between noninvasively reconstructed and the reference pacing sites was 9 (6) mm for the LV epicardial and 12 (5) for the LV epi-endocardial models, 23 (14) for the RV epicardial and 9(4) for the RV epi-endocardial models. There was a significant difference in accuracy for LV epicardial and epi-endocardial models (p=0,007) while no difference for RV models (p=0,039). Conclusions. The study showed sufficient accuracy to use this technology in routine clinical practice for identification of focal arrhythmia sources.