Abstract Ventricular tachycardia (VT) substrate-based ablation has become a gold standard in patients with structural heart disease. Success of VT ablation is related with mortality reduction. Late gadolinium enhancement cardiac magnetic resonance (LGE-CMR) is a powerful technique to assess substrate of VT. Myocardial fibrosis is electrically inert (Core) but it is surrounded by a ‘‘border-zone (BZ)’’ where normal cardiomyocytes intermingle with dense bundles of fibrosis. Slow impulse conduction in the BZ allows for the re-entry circuits leading to VT. Both the presence and extent of LGE have been associated with VT and SCD risk. LGE-CMR tissue characterization can be depicted as pixel signal intensity (PSI) maps and can guide VT ablation. The aim of this study was to analyze possible VT recurrence predictors in a long term follow-up of patients that underwent VT ablation (endo and/or epicardial) related with LGE-CMR PSI maps. We analyzed 234 consecutive patients (age: 63.2 ± 14 years, follow-up: 3.14 years ±1.8) undergoing VT ablation with scar-dechannelling technique at a single center from 2013 to 2018. 110 patients underwent a preprocedural LGE-CMR, and in 94 patients (85,5%) a CMR-aided ablation using the PSI maps was performed. All LGE-CMR images were semi-automatically processed using a dedicated software. PSI-based algorithm was applied to characterize the hyperenhanced area as core or BZ, using fixed threshold of the maximum intensity. A LV 3D shell was obtained and were imported into the navigation system. In the PSI maps, heterogenous tissue channels were defined as a continuous corridor of BZ surrounded by scar core or an anatomic barrier that connects 2 areas of healthy tissue. Results Overall recurrence of VT was 41.8 %. There was ICD shock reduction, from 43,6% to a 28,2% (ICD shocks before ablation 2,23 ± 7,32, after: 1,10 ± 2,92). Left ventricle mass predicted significantly VT recurrence (Mean 168,3 ± 53,3 vs 152,3 ± 46,4 g, HR 1,02 [1,01-1,02], p < 0.001). LGE distribuition was predictive of VT recurrence when a more than 40% of the interventricular septum was involved (62,5% vs 37,8%; HR 1,6 [1,01-1,02]; p = 0,044). No differences in recurrence were found among the patterns of LGE distribution (transmural/epicardial/subendocardial or peculiar segments localizations). The amount of BZ and the total amont of Core + BZ was related with VT recurrence (BZ 26,6 ± 13,9 vs 19,56 ± 9,69 g, HR 1,03 [1,01-1,06], p = 0,012; total Core + BZ 37,1 ± 18,2 vs 29,0 ± 16,3 g, HR 1,02 [1,00-1,04], p = 0,033). Finally VT recurrence was higher in patients with channels with transmural path (66,7% vs 31,4%, HR 3,25 [1,70-6,23], p < 0,001) or midmural channels (54,3% vs 27,6%, HR 2,49 [1,21–5,13], p = 0,013). CMR-aided scar dechanneling is a helpful and feasible technique which could identify patients with high risk of VT recurrence. High left ventricular mass, septal LGE distribution, transmural and midmural heterogeneous tissue channels were predictive factors of post ablation VT recurrence. Abstract Figure. VTchannel & heterogeoneus tissue channel