Previously, we showed that the ganglionated plexi (GP) on the atrium can play a critical role in the initiation and maintenance of atrial fibrillation (AF). We tested the role of the atrial neural network as a substrate for AF without the influence of the GP. In pentobarbital-anesthetized open-chest dogs, two barriers across the left/right atrial appendage (AA) divided the AA into smaller and larger areas of approximately similar size, 2 cmA(2). Electrical stimulation of the superior left and right GP allowed measurement of the greatest percent slowing of the heart rate prior to atrial excitation (n = 7). Acetylcholine (Ach; 1, 10, and 100 mM) was applied to the smaller and then to the larger area. In 22 dogs, the effects on AF duration in response to Ach applied to the atria were tested after GP ablations and atropine applied to the atria. GP function was unchanged by various concentrations of Ach applied to the smaller or larger areas of the atria. However, AF duration was significantly longer for each Ach concentration when applied to the larger versus the smaller area (p a parts per thousand currency signaEuro parts per thousand 0.01). AF was attenuated by GP ablations and atropine, but the differences between small and large areas were maintained. Ach on a larger area of the atria significantly increased the induced AF duration compared to an area half the size without changes in GP function suggesting that recruiting a larger area of the atrial neural network provided more of an AF substrate.