The epicondylar axis is a reliable reference to check the rotation of the femoral implant in total knee prostheses (TKPs). However, during the operation it seems easier to use the posterior condylar axis as a landmark. The angle between these two axes is called the posterior condylar angle (PCA). The aim of this study was to measure the PCA in arthritic knees to assess the reliability of the posterior condylar axis as a reference for the control of the rotation of the femoral implant and to look for correlation with other radiological measurements. This prospective study consisted of 103 arthritic knees (81 varus, 22 valgus) before a TKP had been done in 103 patients (75 women, 28 men). The assessment of the PCA was made by computed tomographic scanning (CT). The HKA, HKS and HKT angles were measured on the pangonogram. The posterior condylar axis was internally rotated with respect to the epicondylar axis. The average value for all the patients was 2.65 degrees degrees with a range from 0 degrees to 7 degrees. The PCA was significantly increased in the valgus knees. There was no correlation between the angles on the pangonogram and the posterior condylar axis. While the preoperative assessment of the PCA by CT scanning is reliable, the results obtained indicate the marked variability in its value. If one wishes to use the posterior condylar axis as a guide for rotation, it is therefore necessary to assess the PCA for each patient using adjustable jigs according to the value obtained. No measurement on standard radiographs allowed an extrapolation of the value of the PCA, and CT scanning seems to be the preferable radiological examination.