AVN Function in Atrial Fibrillation. The irregular ventricular rhythm that accompanies atrial fibrillation (AF) has been explained in terms of concealed conduction within the AV node (AVN). However, the cellular basis of concealed conduction in AF remains poorly understood. Our hypothesis is that electrotonic modulation of AVN propagation by atrial impulses blocked repetitively within the AVN is responsible for changes in function that lead to irregular ventricular rhythms in patients with AF. We have tested this idea using two different simplified computer ionic models of the AVN. The first (“black-box”) model consisted of three cells: one representing the atrium, another one representing the AVN, and a third one representing the ventricle. The black-box model was used to establish the rules of behavior and predictions to be tested in a second, more elaborate model of the AVN. The latter (“nine-cell” model) incorporated a linear array of nine cells separated into three different regions. The first region of two cells represented the atrium; the second region of five cells represented the AV node; and the third region of two cells represented the ventricle. Cells were connected by appropriate coupling resistances. During regular atrial pacing, both models reproduced very closely the frequency dependence of AV conduction and refractoriness seen in patients and experimental animals. In addition, atrial impulses blocked within the AV node led to electrotonic inhibition or facilitation of propagaticm of immediately succeeding impulses. During simulated AF, using the nine-cell model, random variations in the atrial (A-A) interval yielded variations in the ventricular (V-V) interval but there was no scaling, i.e., the V-V intervals were not multiples of the A-A intervals. As such, the model simulated the statistical behavior of the ventricles in patients with AF, including: (1) the ventricular rhythm was random; and (2) the coefficient of variation (standard deviation/mean) of the ventricular rhythm was relatively constant at any given mean V-V interval. Analysis of cell responses revealed that repetitive atrial input at random A-A intervals resulted in complex patterns of concealment within the AVN cells. Consequently, the effects of electrotonic modulation were also random, which resulted in a smearing of the AV conduction curve over A-A intervals that were larger than those predicted for 1:1 AV conduction. Hence, during AF, electrotonic modulation acts in concert with the frequency dependence of AVN conduction to result in complex patterns of ventricular activation. Finally, similarly to what was shown in patients, VVI pacing of the ventricle in the nine-cell model at the appropriate frequency led to blockade of nearly all anterograde (i.e., A-V) impulses. The essential feature here was that the retrograde impulse invading the AVN cells was followed by refractoriness with slow recovery of excitability, setting the stage for electrotonic inhibition of anterograde impulses. Overall, the results provide insight into the cellular mechanisms underlying AVN function and irregular ventricular response during AF.