Electrical stimulation of the vagus nerve has shown efficacy in controlling seizures in experimental models, and early clinical trials have suggested possible benefit in humans. Eleven patients with complex partial seizures were subjected to implantation of vagus nerve stimulators. Electrode contacts embedded in silicone rubber spirals were placed on the left vagus nerve in the low cervical area. A transcutaneously programmable stimulator module was placed in an infraclavicular subcutaneous pocket and connected to the electrode. One patient required replacement of the system due to electrode fracture. Another patient developed delayed ipsilateral vocal-cord paralysis; the technique was then modified to allow more tolerance for postoperative nerve edema. A third patient showed asymptomatic vocal-cord paresis on immediate postoperative laryngoscopy. Vagus nerve stimulation produces transient vocal-cord dysfunction while the current is on. Nine patients were randomly assigned to receive either high- or low-current stimulation, and seizure frequency was recorded. The high-current stimulation group showed a median reduction in seizure frequency of 27.7% compared to the preimplantation baseline, while the low-current stimulation group showed a median increase of 6.3%. This difference approached statistical significance. The entire population then received maximally tolerable stimulation. The high-current stimulation group showed a further 14.3% reduction, while the low-current stimulation group showed a 25.4% reduction compared to the blinded period. The efficacy of vagus nerve stimulation seemed to depend on stimulus parameters, and a cumulative effect was evident. These results are encouraging, and further study of this modality as an adjunct treatment for epilepsy is warranted.