Deep brain stimulation of the thalamus (thalamic DBS) is an established therapy for medically intractable essential tremor and tremor caused by multiple sclerosis. In both disorders, motor disability results from complex interaction between kinetic tremor and accompanying ataxia with voluntary movements. In clinical studies, the efficacy of thalamic DBS has been thoroughly assessed. However, the optimal anatomical target structure for neurostimulation is still debated and has never been analysed in conjunction with objective measurements of the different aspects of motor impairment. In 10 essential tremor and 11 multiple sclerosis patients, we analysed the effect of thalamic DBS through each contact of the quadripolar electrode on the contralateral tremor rating scale, accelerometry and kinematic measures of reach-to-grasp-movements. These measures were correlated with the anatomical position of the stimulating electrode in stereotactic space and in relation to nuclear boundaries derived from intraoperative microrecording. We found a significant impact of the stereotactic z-coordinate of stimulation contacts on the TRS, accelerometry total power and spatial deviation in the deceleration and target period of reach-to-grasp-movements. Most effective contacts clustered within the subthalamic area (STA) covering the posterior Zona incerta and prelemniscal radiation. Stimulation within this region led to a mean reduction of the lateralized tremor rating scale by 15.8 points which was significantly superior to stimulation within the thalamus (P < 0.05, student's t-test). STA stimulation resulted in reduction of the accelerometry total power by 99%, whereas stimulation at the ventral thalamic border (68%) or within the thalamus proper (2.5%) was significantly less effective (P < 0.01). Concomitantly, STA stimulation led to a significantly higher increase of tremor frequency and decrease in EMG synchronization compared to stimulation within the thalamus proper (P < 0.001). In reach-to-grasp movements, STA stimulation reduced the spatial variability of the movement path in the deceleration period by 28.9% and in the target period by 58.4%, whereas stimulation within the thalamus was again significantly less effective (P < 0.05), with a reduction in the deceleration period between 6.5 and 21.8% and in the target period between 1.2 and 11.3%. An analysis of the nuclear boundaries from intraoperative microrecording confirmed the anatomical impression that most effective electrodes were located within the STA. Our data demonstrate a profound effect of deep brain stimulation of the thalamic region on tremor and ataxia in essential tremor and tremor caused by multiple sclerosis. The better efficacy of stimulation within the STA compared to thalamus proper favours the concept of a modulation of cerebello-thalamic projections underlying the improvement of these symptoms.