Background Deep brain electrodes have been used for the last ten years to produce bipolar SEEG-guided radiofrequency thermo-coagulation (SEEG RF-TC). However, this technique is based on empirical knowledge. The aim of this study is threefold: 1) provide in vivo animal data concerning the effect of bipolar RF-TC on brain and its safety 2) assess the parameters of this procedure (current delivery and dipole selection) which produce the most efficient lesion and 3) provide technical guidelines. Methods First we achieved in vivo RF-TC on rabbit brain with several conditions (power delivered and lesioning duration) and analyzed their influence on the lesion produced. Only a difference in terms of volume was found and type of histological lesions was similar whatever the settings were. We then performed multiple RF-TC in vitro on egg albumen first with several parameters of radiofrequency then with different dipole spatial selections. The endpoint was the size of the radiofrequency thermo-lesion produced. Results Using unfixed parameters of radiofrequency current delivery and increasing it until the power delivered by the generator collapsed produced significantly larger lesions (p = 0.008) than other conditions. Concerning the dipole selection, the use of contiguous contacts on electrodes lead to lesions with a higher volume (p = 7.7 x 10-13) than those produced with noncontiguous ones. Conclusion Beside the target selection in SEEG RF-TC, which are summarized based on a literature review, we report the optimal parameters: radiofrequency-current must be increased until the power delivered collapses and dipoles should be constituted by contiguous electrode contacts.