Sixty-five total knee arthroplasties were evaluated by the Knee Society Rad
iological Evaluation System which was developed to encourage uniform report
ing of the results of total knee arthroplasty, All patients were examined b
y three independent experienced radiologists 8.9 years after surgery (range
, 3-16 years) to analyze the interobserver variability. For measurement of
angles, high interobserver correlation was calculated for the prosthetic co
mponent angles and the femorotibial shaft angle, The comparison of the mean
s indicated no significant differences except for the femorotibial shaft an
gle. For measurement of radiolucent lines, interobserver correlation was lo
w for all components. The differences of the means were significantly diffe
rent for all components. The results of interobserver variability of the pa
tellar evaluation revealed high interobserver correlation for the patellar
angle and for patellar subluxation and dislocation evaluation. For assessme
nt of patellar mediolateral and superoinferior displacement, a low interobs
erver correlation was found. For radiographic assessment of total knee arth
roplasty, the measurement of angles, including alpha, beta, femorotibial sh
aft angle, sagittal femoral and tibial component angle, patellar angle, and
patellar subluxation and dislocation evaluation are recommended. The metho
d of assessing radiolucent lines should be reconsidered.