Introduction: Human atria comprise distinct epicardial layers, which can bypass endocardial layers and lead to downstream centrifugal propagation at the “epi-endo” connection. We sought to characterize anatomical substrates, electrophysiological properties, and ablation outcomes of “pseudo-focal” atrial tachycardias (ATs), defined as macroreentrant ATs mimicking focal ATs. Methods and Results: We retrospectively analyzed ATs showing centrifugal propagation with post-pacing intervals (PPIs) after entrainment pacing suggestive of a macroreentry. A total of 26 patients had pseudo-focal ATs consisting of 15 perimitral, 7 roof-dependent, and 5 cavotricuspid isthmus (CTI)-dependent flutters. A low-voltage area was consistently found at the collision site and co-localized with epicardial layers like the: (1) coronary sinus-great cardiac vein bundle (22%); (2) vein of Marshall bundle (15%); (3) Bachmann bundle (22%); (4) septopulmonary bundle (15%); (5) fossa ovalis (7%); and (6) low right atrium (19%). The mean missing tachycardia cycle length (TCL) was 67 ± 29 ms (22%) on the endocardial activation map. PPI was 9 [0-15] ms and 10 [0-20] ms longer than TCL at the breakthrough site and the opposite site, respectively. While feasible in 25 pseudo-focal ATs (93%), termination was better achieved by blocking the anatomical isthmus than ablating the breakthrough site [24/26 (92%) vs. 1/6 (17%); p < 0.001]. Conclusion: Perimitral, roof-dependent, and CTI-dependent flutters with centrifugal propagation are favored by a low-voltage area located at well-identified epicardial bundles. Comprehensive entrainment pacing maneuvers are crucial to distinguish pseudo-focal ATs from true focal ATs. Blocking the anatomical isthmus is a better therapeutic option than ablating the breakthrough site.