Background: Although pulmonary vein isolation (PVI) is the basis of atrial fibrillation (AF) catheter ablation (AFCA), the conditional changes of an anti-AF effect and the wave-dynamic mechanism are poorly understood. Therefore, we compared the effects of PVI based on the ablation circumference, width, and number or locations of gaps with or without amiodarone at the same condition in realistic computational modeling of human AF. Hypothesis: An anatomical level of PVI and characteristics of the PVI gap would affect the anti-AF effects of the AFCA. Methods: We included 50 patients (76.0% persistent AF) who underwent AFCA. Realistic AF modeling reflecting the computed tomography (CT) and the electroanatomical map was performed on each patient. We compared the AF defragmentation rate (DeFR; change to atrial tachycardia (AT) or termination), termination rate (TnR), and changes in the dominant frequency (DF) and the number of phase singularities (PSs) based on the PVI level (antral vs. ostial), PVI width (single vs. 3х width), number and location of 2-mm PVI gaps with or without amiodarone (10mM). We randomly assigned the number and location of the PVI gaps to the AF models. Results: The antral PVI had a significantly higher DeFR (p for trend <0.001) and TnR (p for trend <0.001) than the ostial PVI, PVI gaps, or baseline AF. However, there was no difference depending on PVI width. Among the PVI gaps, the DeFR (p for trend<0.001) decreased as the gap number increased. Additional amiodarone increased DeFR (p=0.003) and TnR (p=0.034) and reduced DF (p<0.001) and the number of PSs (p<0.001) with PVI-gap, especially with a single PVI-gap. Conclusions: In this realistic computational modeling study with integrated patients’ atrial anatomy and electrophysiology, the atrial mass reduction depending on the anatomical PVI level, number of PVI gaps, and additional antiarrhythmic drug significantly affected the AF maintenance mechanisms.