Studies have demonstrated analgesic effects of motor cortex (M1) stimulation for several chronic pain disorders such as neuropathic pain and syndromes involving central sensitization. Central sensitization is an important factor in neuropathic pain, clinically manifested as hyperalgesia and allodynia beyond any apparent injury. We predicted M1 transcranial direct current stimulation (tDCS) would mitigate secondary hyperalgesia, with little or no effect on primary hyperalgesia. We used a capsaicin-heat pain (C-HP) model to elicit heat allodynia and secondary mechanical hyperalgesia in pain-free subjects. In an assessor and subject blind randomized sham-controlled trial, we found anodal M tDCS decreased the intensity and area of pinprick hyperalgesia more than cathodal or sham tDCS with a small to moderate effect size. In contrast, we found no difference among treatments on pain ratings during heat allodynia. These findings confirmed our predictions and support the hypothesis that M1-targeted neuromodulation diminishes central sensitization. To elucidate the mechanism driving analgesia, we repeated application of the C-HP model during anodal, cathodal or sham tDCS in an assessor-blind randomized controlled trial while capturing neurophysiological correlates using functional magnetic resonance imaging (fMRI). We hypothesized M1 anodal tDCS would enhance engagement of a descending pain modulatory (DPM) network in response to mechanical pain compared to cathodal or sham tDCS. Anodal tDCS normalized effects of central sensitization on mechanical pain responses in the DPM network. Anodal tDCS disrupted the normal covariation of mechanical pain processing with subjective pain intensity and blunted the effect of sensitization in primary somatosensory cortex. There were treatment associated differences in functional connectivity (FC) within the DPM network. We found M1 to PAG FC was significantly greater during pain after anodal versus cathodal tDCS. Differences in FC between pain and control states for anodal tDCS included disrupted FC between PAG and sensory regions in the parietal lobe as well as the rostral ventral medulla. No disruptions in FC between control and pain state were found after cathodal or sham stimulation. These results support the hypothesis that analgesia via M1 neuromodulation occurs through modulation of activity in the DPM network even at the earliest stages of therapy.