The purpose of this report is to evaluate the radiographical patellofemoral positioning after total knee arthroplasty (TKA) with different designs in rheumatoid arthritis (RA), and to identify the cause of subluxation of the patella. Between 1982 and 1990, 119 TKA were performed in 87 RA patients, 6 males and 81 females, whose average age at the time of surgery was 57.8 years. The prostheses that were used were Insall-Burstein total condylar knee (I/B II) in 47 knees, Miller-Galante total knee (M/GI) in 50, and Yamamoto Mark IQ total knee (Mark III) in 22. Anteroposterior, lateral and patellar skyline radiographs were made postoperatively to evaluate the results. Patellar prosthesis positioning was evaluated with regard to patellar tilt ( α angle) and lateral patellar shift ratio (PH/AB) . The incidence of postoperative patellar subluxation and dislocation using I/B II, M/G I, and Mark 1ff was 6.4 %, 24 % and 4.5 % respectively. In the M/G I, subluxation or dislocation of patella appeared respective of the Femoro tibial angle. The average postoperative α angle was 2.7 ± 3.10 (I/B II), 10.0±77° (M/G I ) and 4.8±3.7° (Mark III) . The mean of the α angle increased in time in M/G I. The PH/AB was 0.035±0.039 (I/B II), 0.054±0.058 (M/G I) and 0.152±0.030 (Mark III) . The design of the implant appears to be a factor in causing subluxation or dislocation in the M/G I.