From 1971 to 1981, total knee arthroplasty was performed on forty-eight knees in twenty-eight patients with juvenile rheumatoid arthritis at the Robert Breck Brigham (now Brigham and Women's) Hospital. Seventeen of these patients, with twenty-nine knee-replacement arthroplasties, were followed for from two to eleven years (average, five years) and are the basis for this study. The patients' ages at operation ranged from thirteen to thirty-nine years (average, twenty-three years). Six patients had undergone total hip arthroplasty prior to admission for total knee replacement, and five patients had a total hip replacement performed while they were hospitalized for the knee arthroplasty. Thirteen patients (twenty-one knees) had significant preoperative pain but only three (five knees) had severe discomfort. Four patients were unable to walk, three were household walkers, and ten were limited community walkers. Preoperative deformities of the knees ranged from 20 degrees of varus angulation to 35 degrees of valgus angulation. The average preoperative flexion deformity was 23 degrees and the arc of motion averaged 45 degrees. At follow-up, twenty of the twenty-one knees that had been significantly painful preoperatively were completely relieved of discomfort. The average arc of motion increased by 34 degrees, while in all but one knee the angular deformity had been corrected to zero to 10 degrees of valgus angulation. All but one patient became a limited or full community walker. Complications included one late deep infection and one posterior tibial subluxation. Four knees required subsequent resurfacing of the patella for treatment of pain. We now routinely resurface the patella in all patients with juvenile rheumatoid arthritis who have a total knee replacement. To date no prosthesis has required revision for loosening. Radiolucency of one millimeter or less about the prosthesis was noted at follow-up in eight (30 per cent) of the knees. As custom-made components were required in twelve of the twenty-nine knees, it is obvious that preoperative planning is crucial in the treatment of these patients. Our recent experience has shown that the use of preoperative and postoperative serial casts aids greatly in the correction of severe flexion deformity of the knee. Postoperative manipulation was required for twenty-one of the twenty-nine knees. Skeletal immaturity was not an absolute contraindication to surgery. We think that our results, which showed a marked improvement in both knee function and in quality of life, make the short and long-term risks of knee-implant surgery well worth taking in this patient population.