BACKGROUND: Although evidence has indicated a positive effect of transcranial direct current stimulation (tDCS) on reducing pain, few studies have focused on the elderly population with knee osteoarthritis (KOA). OBJECTIVE: To evaluate whether tDCS reduces KOA pain in elderly individuals with a dysfunctional descending pain inhibitory system (DPIS). METHODS: In a double-blind trial, individuals ≥ 60 years with KOA pain and a dysfunctional DPIS, we randomly assigned patients to receive 15 daily sessions of 2 mA tDCS over the primary motor cortex (anode) and contralateral supraorbital area (cathode) (M1-SO) for 20 min or sham tDCS. Change in pain perception indexed by the Brief Pain Inventory (BPI) at the end of intervention was the primary outcome. Secondary outcomes included: disability, quantitative sensory testing, pain pressure threshold and conditioned pain modulation (CPM). Subjects were followed-up for 2 months. RESULTS: Of the 104 enrolled subjects, with mean (SD) age of 73.9 (8.01) years and 88 (84.6%) female, 102 finished the trial. In the intention-to-treat analysis, the active tDCS group had a significantly greater reduction in BPI compared to the sham group (difference, 1.59; 95% CI, 0.95 to 2.23; P < 0.001; Cohen's d, 0.58); and, also a significantly greater improvement in CPM-pressure in the knee (P = 0.01) and CPM-pain in the hand (P = 0.01). These effects were not sustained at follow-up. The intervention was well tolerated, with no severe adverse effects. CONCLUSION: M1-SO tDCS is associated with a moderate effect size in reducing pain in elderly patients with KOA after 15 daily sessions of stimulation. This intervention has also shown to modulate the DPIS.