We reviewed the results of seventy-six revision total knee replacements, performed between 1980 and 1988 on the Knee Service at The Hospital for Special Surgery, in seventy-four patients. Sixty-five patients (sixty-seven knees; 88 percent) had a complete clinical examination and radiographic evaluation, and nine patients (nine knees; 12 percent) were only interviewed by telephone. Survivorship analysis was performed for all patients. The average duration of follow-up was three years and six months (range, two to nine years). Only patients who had had revision of the femoral component or the tibial component, or both, because of aseptic failure were included. The tibial component of all prostheses that were used for revision had a metal backing. Cement was placed on the cut surfaces in the metaphyseal region of the femur and tibia. Fluted diaphyseal intramedullary rods were used in all patients and were not cemented. Metal wedges and augments were used to fill osseous defects when necessary. The average preoperative knee score, according to the rating scale of The Hospital for Special Surgery, was 49 points (range, 0 to 62 points). Postoperatively, the knee score improved to an average of 76 points (range, 0 to 97 points). Of the sixty-seven knees that had complete follow-up, fifty-six (84 percent) had an excellent or good result and five (7 percent) had a fair or poor result. In six (8 percent) of the seventy-six knees, the prosthesis failed, necessitating another revision. Failure was defined as removal or a recommendation for removal of the implant.(ABSTRACT TRUNCATED AT 250 WORDS)