A 20-minute session of 10 Hz repetitive transcranial magnetic stimulation (rTMS) of Brodmann Area (BA) nine of the left dorsolateral prefrontal cortex (DLPFC) can produce analgesic effects on postoperative and laboratory-induced pain. This analgesia is blocked by pretreatment with naloxone, a μ-opioid antagonist. The purpose of this sham-controlled, double-blind, crossover study was to identify the neural circuitry that underlies the analgesic effects of left DLPFC rTMS, and to examine how the function of this circuit, including midbrain and medulla, changes during opioid blockade. Fourteen healthy volunteers were randomized to receive intravenous saline or naloxone immediately before sham and real left DLPFC rTMS on the same experimental visit. One week later, each participant received the novel pretreatment but the same stimulation paradigm. Using short sessions of heat on capsaicin-sensitized skin, hot allodynia was assessed during 3 Tesla functional magnetic resonance imaging (fMRI) scanning at baseline, post-sham rTMS, and post-real rTMS. Data were analyzed using whole-brain voxel-based analysis, as well as time series extractions from anatomically-defined regions of interest representing midbrain and medulla. Consistent with previous findings, real rTMS significantly reduced hot allodynia pain ratings. This analgesia was associated with elevated blood oxygenation-level dependent (BOLD) signal in BAs 9 and 10, and diminished BOLD signal in the anterior cingulate, thalamus, midbrain, and medulla during pain. Naloxone pretreatment largely abolished rTMS-induced analgesia, as well as rTMS-induced attenuation of BOLD signal response to painful stimuli throughout pain processing regions, including midbrain and medulla. These preliminary results suggest that left DLPFC rTMS drives top-down opioidergic analgesia.