A retrospective analysis of revision of customized tumor prostheses of the knee was performed to summarize the surgical technique and reasons for revision, to clarify the occurrence of complications after revision, and the survival and function of the prosthesis. METHODS The clinical data of 33 patients admitted from June 2002 to June 2007 for revision after knee joint customized tumor prosthesis replacement were retrospectively analyzed. There were 17 males and 16 females, aged 16 to 67 years, with an average age of 33.1 years. There were 17 cases of osteosarcoma, 11 cases of giant cell tumor of bone, 2 cases of malignant fibrous histiocytoma of bone, and 1 case each of chondrosarcoma, synovial sarcoma and liposarcoma. Tumor sites: distal femur in 22 cases, proximal tibia in 11 cases. The interval between revision and replacement was from 6 to 180 months, with an average of 45.3 months. Reasons for revision: local recurrence of tumor in 2 cases, periprosthetic infection in 8 cases, aseptic loosening of prosthesis in 7 cases, post-traumatic periprosthetic fracture with loosening of prosthesis in 1 case, fracture of prosthesis shank in 6 cases, failure of prosthesis hinge mechanism in 9 cases, 6 cases of periprosthetic infection, the original prosthesis was taken out for two-stage prosthesis revision; the rest of the patients underwent one-stage revision. The revision prostheses were cemented, and in 2 cases, the length of the residual medullary cavity was shorter than the stem of the revision prosthesis, and a composite prosthesis was implanted with an allogeneic bone segment graft. In the 17 patients who underwent both surgeries in the same hospital, the time for the first replacement surgery was (149.8 ± 40.5) min, and that for the revision surgery was (189.9 ± 43.8) min; the bleeding volume was (605.2 ± 308.0) mL for the replacement surgery, and that for the revision surgery was (834.1 ± 429.9) mL; the differences in the above indexes were all statistically significant (P < 0.05). . The patients were followed up for 12 to 76 months, with a mean of 45.1 months. 2 allogeneic bone grafts reached osseous healing at 1.5 and 2 years postoperatively, respectively. 3 died of lung metastases 12 to 24 months after revision surgery; 3 developed lung metastases and survived with tumors during the follow-up period. Complications occurred in 9 of the 30 surviving patients, with an incidence of 30%; 2 superficial infections, 5 deep infections, and 2 mechanical complications. 7 prostheses failed, with a failure rate of 23.3%. the 5-year survival rate of revision prostheses was 68.6% as analyzed by Kaplan-Meier survival curves. The limb function of the patients was 57.1% ± 10.6% according to the 1993 American Society for Bone Tumor Scoring System (MSTS93) before revision surgery and 73.6% ± 14.4% at 6 months postoperatively, which was statistically significant (P < 0.01). The main reasons for revision of customized tumor prostheses of the knee joint are mechanical problems and infections. Although revision surgery is complicated and has some complications, it can preserve the limb and restore its function in most patients. After upper tibial artificial prosthesis replacement, radiographs showed that the prosthesis was loose and sinking after revision surgery, and joint function was restored