BackgroundIn persistent atrial fibrillation (AF), localized extra-pulmonary vein sources may contribute to arrhythmia recurrences after pulmonary vein isolation. This in-silico study proposes a high-density sequential mapping strategy to localize such sources. MethodCatheter repositioning was guided by repetitive conduction patterns, moving against the prevailing conduction direction (upstream) toward the sources. Sources were found either by locally identifying conduction patterns or by encircling the region harboring them. We simulated source tracking in an in-silico atrial model, comparing random vs. upstream-guided catheter repositioning (with and without encircling). To assess performance in increasing AF complexities, we simulated AF in 3 groups: atria with reentry-anchoring scars, without fibrosis, and with severe endomysial fibrosis. ResultsCompared to random mapping, the upstream-guided approach successfully located sources more often (anchored reentries: 70.6% vs. 10.6%; no fibrosis: 87.9% vs. 22.1%; with fibrosis: 95.0% vs. 60.9% of tracking procedures, all p<0.001), using fewer steps (median [IQR]: 11 [7;23] vs. 26 [13;35]; 10 [6;19] vs. 19 [10;27]; 11 [7;19] vs. 16 [8;30], respectively, all p<0.05). Adding source encircling increased source detection (98.1 %, 100 %, and 99.5 %, all p<0.01 vs. local detection only), reducing required steps (9 [6;12], 8 [6;12], and 9 [6;13], all p<0.05). In some cases (11.9 %, 17.1 %, and 10.5 % of procedures), the algorithm encircled regions >15 mm from the source. ConclusionMoving mapping catheters upstream improves source detection efficiency, even in the presence of severe fibrosis. Encircling sources may help find regions of interest in fewer steps.