The models of cardiac electrical activity are theoretical schemes of electro physiological phenomena based on mathematical formulations and are part of efforts to facilitate understanding and prediction of their behavior in normal and pathological situations. Mathematical modeling and virtual anatomical structures, coupled with computer simulation to help analyze and understand in more detail the origin of reentry leading to atrial arrhythmias of electrical origin, because the inherent complexity of this phenomenon is very difficult study using only experimentally. This work is based on the development of two tree-dimensional models of human atria, a model of physiological dimensions and one with left atrial dilatation (structural remodeling). Both models were incorporated a realistic fiber orientation and anisotropic conduction. Electrical remodeling and the remodeling of gap junctions, occasioned by chronic atrial fibrillation episodes were simulated. Thus, the study includes four different models: 1) a model in physiological conditions, with anatomical and electric normal features 2) a model of normal anatomical features, but in terms of electrical remodeling, 3) a model with dilatation and electrical remodeling, and 4) a model with dilation and electrical remodeling and gap-junctions remodeling. The models reproduced the spread of action potential in physiological and pathological situations. By simulation, we studied the effect of electrical remodeling, structural remodeling (dilatation) and gap-junctions remodeling on vulnerability to reentry. In general, the results show as electrical remodeling favors the generation of reentry. The increased vulnerability for atrial reentry is even higher in dilated atria with gap-junctions remodeled. We found also that the location of the ectopic foci does vary the vulnerability significantly. The focal activity analysis found a strong impact of the anatomy to maintain permanent reentry. The occurrence of figure-eight reentry, spiral-and macro-reentry were observed in pathological models.