Purpose: Experimental studies showed that local capture of atrial fibrillation (AF) by rapid pacing was possible in humans. However, contradictory observations were reported on its effect at distant atrial sites. The present model-based study investigated the effect of rapid pacing on the paced and the non-paced atria. Methods: A biophysical model of AF based on a geometry from human MRI and a membrane kinetics model was used. Rapid AF pacing was applied during 30s in right atrial (RA) free wall or left atrial (LA) appendage at optimal pacing cycle lengths based on previous studies (RA:76 ms, LA:77ms). The pacing effect was characterized by measuring 256 electrograms evenly located in RA and LA, from which the following values were computed: AF cycle length (AFCL), number of wavefronts (#WF), percentage of excited tissue (ET) and AF organization index (OI) assessing spectral regularity and ranging from 0 to 1. Results: Pacing successfully induced local AF capture in the paced atrium with an AFCL close to the pacing cycle length. Local capture was accompanied by a significant (p<0.001) reduction in #WF and increase in ET and OI in the paced atrium. In the non-paced atrium, AFCL only slightly decreased while the effect on #WF was the opposite to what was observed in the paced atrium (increase) and ET did not change. Interestingly, the effect on OI was different when pacing from the LA (decrease in the non-paced atrium compared to no pacing) or the RA (increase). Conclusions: The effect of AF rapid pacing on the non-paced atrium showed an acceleration of AF with an increased number of wavefronts. However, the effect on AF organization was dependent on the pacing site, which could explain the contradictory results reported in humans.