Introduction: We recently observed in an acute ovine model of pacing-induced atrial tachycardia (PIAT) that rapid atrial capture and fibrillatory conduction were prevented by 2:1 capture mediated by decreased excitability. We hypothesize that long-term intermittent PIAT mimicking pulmonary vein tachycardia promotes rapid atrial capture beyond the ERP measured in the baseline state, fibrillatory conduction and sustained atrial fibrillation (AF) by decreasing activation recovery interval (ARI), a surrogate of action potential duration (APD). Methods: We specifically developed a chronic ovine model of PIAT using two pacemakers (PM) each with a right atrial (RA) lead separated by ~2cm. The 1st PM (Vitatron T70) was used to record a broadband unipolar RA EGM (800 Hz, 0.4 Hz high pass filter). The 2nd was used to deliver PIAT during electrophysiological protocols at decremental pacing CL (400 beats, from 400 to 110ms) and long term intermittent RA burst pacing to promote electro-anatomical remodelling (5s of burst followed by 2s of sinus rhythm) until onset of sustained AF. ARIs, defined as the time difference between the peaks of the atrial depolarisation and repolarisation waves, were averaged over 20 consecutive beats at each pacing CL and their restitution kinetics as a function of the pacing CL were compared before (i.e. in the baseline state), during and after remodelling (i.e. the week before sustained AF occurred) using multiple-way ANOVA. A p<0.05 was considered significant. Logarithmic regression was used to fit the ARI-pacing CL relationship (y=a+b ln(x)). Results: Intermittent PIAT induced sustained AF on average after 6.0±3.5 weeks of burst pacing (n=8 sheep). The figure shows that ARI measured at 400 ms pacing CL before remodelling (122±20 ms) was significantly decreased by intermittent PIAT during and after remodelling (99±16 ms and 98±20 ms respectively, p<0.05). Moreover, the rate-dependence of ARI was significantly altered by intermittent PIAT from a steep kinetics (b=34.1, p<0.01) before remodelling to a nearly flat relationship during (b=10.9, p=NS) and after remodelling (b=10.9, p=NS). Importantly, there was no difference in atrial structure and fibrosis as assessed by light microscopy between remodelled sheep and a set (n=4) of unremodelled control sheep. Conclusions: Using standard pacemaker technology, atrial ARIs as a surrogate of APDs were successfully measured in vivo during the electrical remodelling process leading to AF. The facilitation of AF by pacing-induced atrial tachycardia mimicking salvos from pulmonary veins is heralded by a significant shortening of ARI, with loss of rate-dependence in the remodelled as compared to the unremodelled state.