Generally, exposure and the operative technique are common difficulty enconuntered in the total knee replacement for old infectious arthritis. Also the postoperative range of motion is poor due to deformity and excessive soft tissue contracture around knee. The surgeon will often be disappointed at the poor flexion after prosthetic implantation, relocation of the patella and wound closure. In the period from August 1986 to August 1990, twenty-two patients with a well-defined history of old septic or tuberculous infection underwent total knee arthroplasty. We analysed eighteen patients who could be followed up more than one year. The follow-up has been five years and two months in the longest case and one year in the shortest case, thirty months on an average. The knee infection had been tuberculous in eleven cases, septic or pyogenic in seven cases. The age of onset of infection varied from seven year to fifty-nine year old. The age of patients at the time of total knee arthroplasty ranged from twenty year to sixty year old, the average being thirty-three years and six months. The quiescent period between effective care of infectious arthritis and the time of total knee arthroplasty had varied from less than one year to more than thirty years. During the procedure there needed frequently soft tissue release such as lengthening of quadriceps tendon, proximal transfer of patellar tendon or tibial tubercle. The results analysed are as follows:1. There were twelve knees with complete ankylosis preoperatively. The average preoperative range of motion of the remaining six knees was fifty-four degrees (range 5-130 degrees). The postoperative range of motion of twelve knees with complete ankylosis and the remaining six knees was sixty-six degrees (range 10-120 degrees) and one hundred and three degrees (75-140 degrees) respectively. 2. Knee Rating Score of Hospital for Special surgery was improved from 62.5 preoperatively to 85 postoperatively. 3. Radiolucent line was observed in one knee with less than 1 mm width in three years and four months postoperatively, but the patient had no pain. 4. There were one superficial infection and one deep infection. The deep infection was hematogenous infection caused by β-hemolytic streptococcus. In two knees who had unsatisfactory range of motion we performed intra-articular fibro-arthrolysis under arthroscopic control.