Background: The location of the optimal target for deep brain stimulation (DBS) of the subthalamic nucleus (STN) remains controversial. Electrode impedance affects tissue activation by DBS and has been found to vary by contact number, but no studies have examined association between impedance and anatomic location. Objectives: To evaluate the relationship between electrode impedance and anatomic contact location, and to assess the clinical significance of impedance. Methods: We gathered retrospective impedance data from 101 electrodes in 73 patients with Parkinson's disease. We determined contact location using microelectrode recording (MER) and high-field 7T MRI, and assessed the relationship between impedance and contact location. Results: For contact location as assessed via MER, impedance was significantly higher for contacts in STN, at baseline (111 Ω vs STN border, p=0.03; 169 Ω vs white matter, p<0.001) and over time (90 Ω vs STN border, p<0.001; 54 Ω vs white matter, p<0.001). Over time, impedance was lowest in contacts situated at STN border (p=0.03). Impedance did not vary by contact location as assessed via imaging. Location determination was 75% consistent between MER and imaging. Impedance was inversely related to absolute symptom reduction during stimulation (-2.5 motor portion of the Unified Parkinson's Disease Rating Scale (mUPDRS) points per 1000 Ω, p=0.01). Conclusions: In the vicinity of DBS electrodes chronically implanted in STN, impedance is lower at the rostral STN border and in white matter, than in STN. This finding suggests that current reaches white matter fibres more readily than neuronal cell bodies in STN, which may help explain anatomic variation in stimulation efficacy.