Background This study assessed the effect of blockading neural transmission in the ganglionated plexi by injecting lidocaine into fat pads in the vagal nerve stimulation canine model and patients with persistent atrial fibrillation ( AF ). Methods and Results An efficacy test of lidocaine injection was performed in 7 canines. During vagal nerve stimulation, AF was sustained for >5 minutes. The lidocaine was injected into ganglionated plexi during sinus rhythm and reinduction of AF was attempted. Six patients with persistent AF were studied at open heart surgery. Lidocaine was injected into ganglionated plexi. Atrial electrograms were recorded from 96 epicardial electrodes covering Bachmann's bundle and atrial appendages. In the canine vagal nerve stimulation AF model, AF was not inducible in 4 of 7 after lidocaine injection. In patients with persistent AF , during baseline AF , there was a left atrium ( LA )-to-right atrium ( RA ) frequency gradient ( LA , mean cycle length [ CL ] 175±17 ms; RA , mean CL 192±17 ms; P<0.01). After lidocaine injection, AF persisted in all patients, and the LA -to- RA frequency gradient disappeared ( LA , mean CL 186±13 ms; RA , mean CL 199±23 ms; P=0.08). Comparison of mean CL s before and after lidocaine demonstrated prolongation of LA CL s ( P<0.05) with no effect on RA CL s. Conclusions In the canine vagal nerve stimulation AF model, lidocaine injection decreased inducibility of AF . In patients with persistent AF , atrial electrograms from the LA had shorter CL s than RA , indicating an LA -to- RA frequency gradient. Lidocaine injection significantly prolonged only LA CL s, explaining disappearance of the LA -to- RA frequency gradient. The mechanism of localized atrial electrogram CL prolongation in patients with persistent AF is uncertain.