Segmental thoracic neuropathic pain (NeuP) remains particularly difficult to treat. Sensory ganglionectomy was reported to alleviate NeuP. The experience with thoracic ganglionectomy, however, is very limited. Here, we report the results of a prospective pilot study in patients with incapacitating segmental thoracic NeuP treated by selective ganglionectomy. Seven patients were included suffering from refractory NeuP scoring 8 or more on a visual analogue scale (VAS). Every patient had test anaesthesia prior to surgery yielding more than 50% pain relief. The spinal ganglion was excised completely via an extraforaminal approach. Mean preoperative VAS scores were 9.1 (maximum pain); 5.4 (minimum pain); 7.9 (pain on average); 6.9 (pain at the time of presentation); and 7.4 (allodynia). Early post-operatively, there was a marked improvement of mean scores: 1.7; 0.7; 1.2; 1.0; and 0.7, respectively. One patient developed a mild transient hemihypaesthesia. In three patients, substantial pain occurred in a formerly unaffected dermatome within 1 year. Two of these patients had significant pain relief by a second operation. At the time of last follow-up at a mean of 24 months after the first procedure, mean VAS scores were 6.3; 2.1; 4.3; 4.0; and 1.3. Overall, medication was reduced. The patients rated their outcome as excellent (1), good (2), fair (2) and nil (2) with best improvement for allodynia. Selective thoracic ganglionectomy is a safe and partially effective procedure in selected patients albeit there may be partial recurrence of pain. Recurrent pain may affect dermatomes that were not involved initially.