The impact of atrial dispersion of refractoriness (Disp_A) in the inducibility and maintenance of atrial fibrillation (AF) has not been fully resolved. Aim: To study the Disp_A and the vulnerability (A_Vuln) for the induction of self-limited (b 60 s) and sustained episodes of AF. Methods and results: Forty-seven patients with paroxysmal AF (PAF): 29 patients without structural heart disease and 18 with hypertensive heart disease. Atrial effective refractory period (ERP) was assessed at five sites -right atrial appendage and low lateral right atrium, high interatrial septum, proximal and distal coronary sinus. We compared three groups: group A -AF not inducible (n = 13); group B -AF inducible, self-limited (n = 18); group C -AF inducible, sustained (n = 16). Age, lone AF, hypertension, left atrial and left ventricular (LV) dimensions, LV systolic function, duration of AF history, atrial flutter/tachycardia, previous antiarrhythmics, and Disp_A were analysed with logistic regression to determine association with A_Vuln for AF inducibility. The ERP at different sites showed no differences among the groups. Group A had a lower Disp_A compared to group B (47 ± 20 ms vs 82 ± 65 ms; p = 0.002), and when compared to group C (47 ± 20 ms vs 80 ± 55 ms; p = 0.008). There was no significant difference in Disp_A between groups B and C. By means of multivariate regression analysis, the only predictor of A_Vuln was Disp_A (p = 0.04). Conclusion: In patients with PAF, increased Disp_A represents an electrophysiological marker of A_Vuln. Inducibility of both self-limited and sustained episodes of AF is associated with similar values of Disp_A. These findings suggest that the maintenance of AF is influenced by additional factors.