Objectives: Rapid pacing of atrial fibrillation (AF) can induce local atrial capture. The present model-based study investigated the impact of atrial tissue conduction velocity on AF capture ability during rapid septal pacing. Methods: The AF model combined a membrane kinetics model with geometry based on computed tomography of AF patients. Conduction velocity was varied ±20% over a baseline AF model based on multiple reentrant wavelets. Rapid pacing of AF was applied from the septum for 50s with pacing cycle length (PCL) computed as percent of mean AFCL. Analysis of 24 electrode pairs evenly distributed on the atrial surface yielded percentage of captured tissue (CL within ±5% of PCL). Capture window was the range of PCL with capture > 50%. Reentrant wavelets quantity (#W) was computed before and during pacing. Optimal PCL was leading to the highest capture. Results were averaged on 10 AF simulations. Results: AFCL did not change significantly with conduction velocity, and optimal PCL was comparable for the 3 velocity values. An increase/decrease in velocity reduced/extended the capture window. The maximum capture was obtained with a decreased conduction velocity even if #W was higher prior to pacing. Conclusions: Changes in atrial tissue conduction properties produced significant differences in rapid pacing outcomes, suggesting that different types of AF may respond differently to therapeutic pacing. Optimal capture parameters depended on AF dynamics but not AFCL.