BACKGROUND: Computational models of atrial fibrillation (AF) help clarify arrhythmia mechanisms, but the contribution of left atrial (LA) epicardial adipose tissue (EAT) is rarely included. OBJECTIVE: To assess whether incorporating EAT-associated effects into fibrosis-based atrial models alters reentrant arrhythmia inducibility, and to characterize the pro-arrhythmic properties of EAT-affected tissue. METHODS: LA fibrosis and EAT were quantified by LGE and Dixon MRI in 19 patients with persistent AF. For each patient, three anatomically identical image-personalized models were built (fibrosis only; fibrosis plus half-epicardial EAT effects, 0.5 mm; fibrosis plus full-epicardial EAT effects, 1.0 mm), yielding 855 simulations. In EAT regions, inward rectifier K+ current was reduced by 35%, intracellular K+ by 50%, and conduction velocity by 25%. RESULTS: Sustained reentry occurred in 370 simulations (43.3%), of which 306 had a functional (rotor-driven) component. Functional reentry inducibility rose from 22.8% (fibrosis only) to 41.4% (half-EAT) and 43.2% (full-EAT). At least one functional reentry was inducible in 12/19 fibrosis-only models, 18/19 half-EAT models, and 19/19 full-EAT models. Median inducible simulations per patient increased from 4 to 6 (p=0.004) and 7 (p=0.003), without significant difference between half- and full-EAT (p=0.895). Fibrosis/EAT overlap regions comprised only 2.1% of the LA wall but were the most arrhythmogenic, with a median 65.5% of their volume harboring reentry versus 8.8% (non-substrate), 37.3% (fibrosis), and 42.8% (EAT) (all p<0.001). In an exploratory analysis of PVI-only ablation patients (n=13), fibrosis-EAT overlap outside the PVI region was associated with 1-year recurrence (log-rank p=0.007). CONCLUSION: Adding EAT effects to fibrosis-based models increased sustained functional reentry, with transmural depth contributing only modestly. Reentry anchored preferentially where fibrosis and EAT effects coexisted.