Despite the introduction of new antiepileptic drugs and advances in the surgical treatment of epilepsy, an important group of patients still remains uncontrolled by any of these methods. The relatively recent introduction of vagus nerve stimulation is yet another possible treatment for refractory epilepsy. This safe, simple, and adjustable technique reduces the number of seizures and multiple publications support its increasing efficacy and effectiveness, with few adverse effects. The goal of our study is to determine the efficacy of this procedure and the factors predicting a response, particularly in the presence of a temporal lobe discharge on the video electroencephalogram (video-EEG) and magnetic resonance imaging (MRI) lesions. We undertook a retrospective study of all the patients with refractory epilepsy who underwent implantation of a vagus nerve stimulator between 2003 and 2009, and with a minimum follow-up of 6 months. The statistical analysis was done with SPSS for Windows. The stimulator was implanted in 40 patients, of whom 38 had a minimum follow-up of 6 months. In one patient, the device had to be removed due to infection, so the series comprised 37 patients. These were divided into different groups, according to the epidemiologic, clinical, radiologic, and electroencephalographic data. In addition, an analysis of the response was performed. The efficacy of the procedure was established according to the reduction in the mean seizure frequency. The baseline value of these seizures was 80.97 ± 143.59, falling to 37 ± 82.51 at the last revision. The response rate (reduction in seizures ≥ 50 %) at 6 months was 51.4 %, with 62.2 % of the patients showing this reduction at the last evaluation. Significant differences in the response were seen for the variables: baseline frequency of seizures, temporal lobe discharge on VideoEEG and MRI lesions. The mean time to response was 10 months in patients with lower rate of seizures versus 25 months of those with the higher rate (p = 0.024), and the response at 6 months was higher (p = 0.05). Patients with temporal lobe discharge alone or in combination with discharges over other regions had a mean time to response of 11 months versus 26 months in those without temporal discharge (p = 0.037). In the analysis of the MRI, we had seen that at the last revision, 82.4 % of the patients with lesion had achieved response versus 45 % without lesion (p = 0.02). Vagus nerve stimulation reduces the frequency of seizures. A temporal lobe discharge on the video-EEG is an indicator of an early response and the presence of an MRI lesion indicates a late response. Patients with fewer rates of seizures have a better prognosis.