The clinical results after knee revision arthroplasty using the Total Condylar III (TC III) prosthesis were reviewed in 36 patients (36 knees). Patients were assessed by physical examination, completion of a 100-point, modified Hospital for Special Surgery knee rating form, and roentgenographic evaluation. There were 27 women and nine men, with a mean age of 68 years. Mean follow-up period was 45 months (range, 24-84 months). Reason for revision was sepsis in 15 patients and loosening or instability in 21 patients. Twenty were Category C patients with multiple joint involvement or significant medical infirmity. Preoperative knee scores ranged from 11-72 points (mean, 36 points). Postoperative scores ranged from 40-99 points (mean, 77 points). Clinical results were graded according to Insall. Eleven patients were graded excellent, 14 patients good, six fair, four poor, and there was one failure. Pain scores improved from a mean preoperative score of 16 points to 43 points postoperatively (50 points possible). However, function scores improved less dramatically. Range of motion based on maximum flexion averaged 98 degrees postoperatively. Varus tibial placement occurred in 50% (range, 1 degree-9 degrees). Slight flexion of the femur was seen in 73%. Lucent lines occurred in 60%, with 16% of the patients showing progressive tibial lucencies. The TC III prosthesis for revision arthroplasty is effective for salvaging unstable knees. Since the natural history of tibial radiolucency is not clear, less-constrained devices should probably be used when possible.