Surgical options for the failed total knee arthroplasty include resection arthroplasty, arthrodesis, and total knee revision. Resection arthroplasty has not been a satisfactory alternative for revision of the failed total knee in a rheumatoid patient population. The arthrodesis experience has been associated with a high incidence of pseudarthrosis (30%), significant shortening, a high complication rate, and poor functional results. Sixty-five failed noninfected total knee arthroplasties were revised to metal-to-plastic total knee arthroplasty. The most common mechanism of failure was loosening (57%). The average follow-up period following revision was 33 months (range, 24-60 months). Fifty-three of the knees were improved at last follow-up examination, compared with the prerevision score. The average postoperative score in the revision series was 80 points. Eight knees have required further surgery. Total knee revision of the failed noninfected total knee arthroplasty is a successful and preferable alternative to knee arthrodesis. Knee revision surgery is technically demanding and requires the availability of multiple components of varying constraint, including custom-made components.