Our purpose was to determine the mechanism which allows the maximum knee flexion in vivo after a posterior-cruciate-ligament (PCL)-retaining total knee arthroplasty. Using three-dimensional computer-aided design videofluoroscopy of deep squatting in 29 patients, we determined that in 72% of knees, direct impingement of the tibial insert posteriorly against the back of the femur was the factor responsible for blocking further flexion. In view of this finding we defined a new parameter termed the 'posterior condylar offset'. In 150 consecutive arthroplasties of the knee, the magnitude of posterior condylar offset was found to correlate with the final range of flexion.