Patients with advanced stages of essential tremor frequently exhibit tandem gait ataxia with impaired balance control and imprecise foot placement, resembling patients with a cerebellar deficit. Thalamic deep brain stimulation, a surgical therapy for otherwise intractable cases, has been shown to improve tremor, but its impact on cerebellar-like gait difficulties remains to be elucidated. Eleven patients affected by essential tremor (five females; age 69.8 ± 3.9 years; disease duration 24.4 ± 11.2 years; follow-up after surgery 24.7 ± 20.3 months) were evaluated during the following conditions: stimulation off, stimulation on and supra-therapeutic stimulation. Ten age-matched healthy controls served as the comparison group. Locomotion by patients and controls was assessed with (i) overground gait and tandem gait; (ii) balance-assisted treadmill tandem gait and (iii) unassisted treadmill gait. The two treadmill paradigms were kinematically analysed using a 3D opto-electronic motion analysis system. Established clinical and kinesiological measures of ataxia were computed. During stimulation off, the patients exhibited ataxia in all assessment paradigms, which improved during stimulation on and worsened again during supra-therapeutic stimulation. During over ground tandem gait, patients had more missteps and slower gait velocities during stimulation off and supra-therapeutic stimulation than during stimulation on. During balance-assisted tandem gait, stimulation on reduced the temporospatial variability in foot trajectories to nearly normal values, while highly variable (ataxic) foot trajectories were observed during stimulation off and supra-therapeutic stimulation. During unassisted treadmill gait, stimulation on improved gait stability compared with stimulation off and supra-therapeutic stimulation, as demonstrated by increased gait velocity and ankle rotation. These improvements in ataxia were not a function of reduced tremor in the lower limbs or torso. In conclusion, we demonstrate the impact of thalamic stimulation on gait ataxia in patients with essential tremor with improvement by stimulation on and deterioration by supra-therapeutic stimulation, despite continued control of tremor. Thus, cerebellar dysfunction in these patients can be differentially modulated with optimal versus supra-therapeutic stimulation. The cerebellar movement disorder of essential tremor is due to a typical cerebellar deficit, not to trembling extremities. We hypothesize that deep brain stimulation affects two major regulating circuits: the cortico-thalamo-cortical loop for tremor reduction and the cerebello-thalamo-cortical pathway for ataxia reduction (stimulation on) and ataxia induction (supra-therapeutic stimulation).