Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is increasingly used to treat advanced Parkinson's disease (PD). The optimal method for targeting the STN before implanting the definitive DBS electrode is still a matter of debates. Beside methods of direct visualization of the nucleus based on stereotactic magnetic resonance imaging (MRI), the most often used technique for targeting STN consists in recording single-cell activity along exploratory tracks of 10-15mm in length, centered on the theoretical or MRI-defined target coordinates. Single-unit recordings with a microelectrode present various drawbacks. They are time-consuming if correctly performed and a single-cell precision is probably superfluous, taking into account the size of the implanted electrode. In this study, we present an original method of recording and quantification of a multi-unit signal recorded intraoperatively with a semi-microelectrode for targeting the STN. Twelve patients with advanced PD have been included and assessed clinically before and one year after bilateral STN-DBS electrode implantation guided by multi-unit electrophysiological recordings. After one year of chronic stimulation, all patients showed a marked clinical improvement. The motor score of the unified Parkinson's disease rating scale decreased by more than 57% and the required levodopa-equivalent daily dose by 59.5% in on-stimulation off-medication condition compared to off-stimulation off-medication condition. The accuracy of STN-DBS lead placement was confirmed on postoperative computed tomography (CT) scans, which were fused to preoperative T2-weighted MRI. The boundaries of the STN were easily determined by an increase in multi-unit signal amplitude, which was observed on average from 0.492mm below the rostral border of the STN down to 0.325mm above its caudal border. Signal amplitude significantly increased at the both rostral and caudal STN margins (P<0.05) and the level of neuronal activity easily distinguished inside from outside the nucleus. This study showed that STN boundaries could be adequately determined on the basis of intraoperative multi-unit recording with a semi-microelectrode. The accuracy of our method used for positioning DBS electrodes into the STN was confirmed both on CT-MRI fusion images and on the rate of therapeutic efficacy.