PURPOSE: To determin whether radiographic assessment of patella position abnormalities (such as patella altaor baja) using the 'patellar tendon : patella ratio' (Insall & Salvati's method) can be applied to flexion andextension sagittal MR images of the knee MATERIALS AND METHODS: Both flexion (55-90 degrees, mean 62 degrees) and extensionT2-weighted sagittal images of 58 knees (54 patients;age:19-72 years:M:F=40:18 were obtained and reviewed by meansof routine lateral radiographs (flexion of knees:30-60 degrees). Patellar tendon and diagonal length were measured onroutine radiographs and on flexion and extension MR images;length was measured by sagittal T2-weighted MR imagingfrom the inner aspect of patellar insertion to the inner asspect of tibial insertion, an approach whichdemonstrated the even thickness of the patellar tendon and the greastest diameter of the patella. Usingcorrelation analysis, the ratio of patellar tendon to diagonal length was compared between examination methods andbetween the non-wavy and wavy forms of patellar tendon. RESULTS: The mean length of the respectirely, 'patella,patellar tendon and patellar tendon to patella ratio' were 4.15+/-0.38, 4.04+/-0.49 and 0.9+/-0.12, respectively, onroutine radiographs; 4.32+/-0.36, 3.8+/-0.47 and 0.89+/- 0.12, respectively, on flexion MR images;and 4.3+/-0.36,3.93+/-0.44 and 0.92+/-0.12, respectively on extension MR images. Between these three different modes of assessment,a high degree of correlation of mean patella length (r=0.89-0.92) and of mean patellar tendon length (r=0.71=0.85)were found. In addition, correlation of 'patella to patellar tendon ratios' was excellent : radiographs andflexion MRimages(r=0.76);radiographs and extension MR images(r=0.76);flexion and extension MR images(r=0.84).According to the form of patellar tendon, radiographs correlated better with extension MR images (r-0.71) thanwith flexion images (r=0.62) in the non-wavy form of patellar tendon, whereas in the wavy form, correlation withflexion images (r=0.83) was better than with extension images(r=0.78). CONCLUSION: MR asessment of 'patellartendon to patella ratio's correlated well with radiographic assessment, and the latter can therefore be used forthe evaluation of abnormal patellar position. Radiographs correlated better with extension MR images in thenon-wavy form of patellar tendon, whereas in the wavy form, correlation with flexion images was better.