Background: The rotor hypothesis has gained favor in recent years for treating chronic atrial fibrillation (CAF) trough ablation. However, catheter ablation can cause complications in patients. Rotor modulation via pharmacologic intervention can reduces the ablation risks. Dofetilide is an antiarrhythmic drug indicated for CAF, but it may trigger the ventricular tachycardia torsade de pointes by the APD and QT interval prolongation. Therefore, dofetilide should be delivered in atrial specific targets without affecting the ventricles. Aim: To evaluate the effect of the localized release of dofetilide on rotors generated in CAF using a computational model. Methods: A computational model of two-dimensional atrial tissue was implemented using the Courtemanche human atrial cell model under CAF conditions. The dofetilide model consists in blocking the ionic currents IKr and IKACh using Hill’s equation. A CAF episode, sustained by a rotor, is simulated by applying the S1-S2 cross-field stimulation protocol. Phase map analysis was applied to detect the rotor tip. Different concentrations of dofetilide are applied to the whole tissue and on geometrical patterns based on the rotor tip location. Results: When we applied dofetilide (0-500 uM) on the whole tissue, the rotor becomes more stable, independently of the concentration. Implementing a ring pattern of dofetilide surrounding the tip, the rotor becomes a macro-reentry. Finally, with dofetilide applied specifically in a rectangle shaped pattern, the rotor activity ends at different concentrations with a minimum of 5 uM. Conclusion: Non-specific application of dofetilide can benefit the progression of CAF, which is consistent with clinical observations of this drug. We show that the CAF episode ended through localized dofetilide liberation. These results could have therapeutic implications in novel treatments of CAF.