Despite the high percentage of patients who are satisfied completely with the results of total knee arthroplasty, there is a small group who remain disabled because of persistent neuroma pain. Recently, a better understanding of the innervation of the skin and capsule around the human knee joint has provided a theoretical basis for denervation in this group of patients. Fifteen patients were identified who had persistent or worse knee pain for > 6 months after total knee arthroplasty. In each patient, component loosening, malalignment, knee instability, and infection had been excluded systematically as a source of pain. Pain was evaluated preoperatively with the Knee Society Functional Score Questionnaire and a visual analog scale. To be selected for surgery, each patient must have had a reduction by 5 points on the visual analogue scale for pain after undergoing selective nerve blocks. Postoperative assessment was done by a team that did not include the surgeon who did the denervation. The technique for selective knee denervation is described in detail. All 15 patients had at least 1 of the nerves to the knee selectively denervated (45 nerves in 15 patients). All patients reported subjective improvement in the immediate postoperative period. This improvement was maintained at a mean followup of 12 months (range, 6-16 months). Selective knee denervation is indicated in the management of intractable knee pain of neuroma origin after total knee arthroplasty.