High-frequency repetitive transcranial magnetic stimulation (rTMS) of the primary motor cortex for neuropathic pain has been shown to be effective, according to systematic reviews and therapeutic guidelines. However, our large, rigorous, investigator-initiated, registration-directed clinical trial failed to show a positive primary outcome, and its subgroup analysis suggested that the analgesic effect varied according to the site of pain. The aim of this study was to investigate the differences in analgesic effects of rTMS for neuropathic pain between different pain sites by reviewing our previous clinical trials. We included three clinical trials in this mini meta-analysis: a multicenter randomized controlled trial at seven hospitals (N = 64), an investigator-initiated registration-directed clinical trial at three hospitals (N = 142), and an exploratory clinical trial examining different stimulation parameters (N = 22). The primary efficacy endpoint (change in pain scale) was extracted for each patient group with pain in the face, upper limb, or lower limb, and a meta-analysis of the efficacy of active rTMS against sham stimulation was performed. Standardized mean difference (SMD) with 95% confidence interval (CI) was calculated for pain change using a random-effects model. The analgesic effect of rTMS for upper limb pain was favorable (SMD = −0.45, 95% CI: −0.77 to −0.13). In contrast, rTMS did not produce significant pain relief on lower limb pain (SMD = 0.04, 95% CI: −0.33 to 0.41) or face (SMD = −0.24, 95% CI: −1.59 to 1.12). In conclusion, these findings suggest that rTMS provides analgesic effects in patients with neuropathic pain in the upper limb, but not in the lower limb or face, under the conditions of previous clinical trials. Owing to the main limitation of small number of studies included, many aspects should be clarified by further research and high-quality studies in these patients.