Deep knee flexion is important to proper function for some activities and in some cultures, although there are large posterior forces during high knee flexion. Most of what we know about posterior restraint and stability, however, has not been determined from deep flexion and without distinguishing motion in the medial and lateral compartments. We therefore evaluated (1) the difference in posterior displacement between the medial and lateral compartments at a commonly used flexion angle of 90 degrees ; (2) that of deeply flexed knees at 135 degrees ; and (3) the difference in kinematics in the medial and lateral compartments. We analyzed posterior stability in 21 normal knees using interventional open magnetic resonance imaging (MRI) system. When manual posterior stress was applied, the posterior displacements of the tibia were 0.6 mm/2.1 mm (medial/lateral) at 90 degrees and 0.6 mm/3.6 mm at 135 degrees . The posterior aspect of the femoral medial condyle moved 7.5 mm anteriorly with knee flexion, whereas the lateral condyle moved 1.3 mm anteriorly. The contact point of the lateral compartment moved 9.2 mm posteriorly with knee flexion, whereas the contact point of the medial compartment moved 2.3 mm anteriorly. Posterior displacement was larger in the lateral compartment at both flexion angles with manual posterior stress. As the knees flexed from 90 degrees to 135 degrees , posterior displacement became larger in the lateral compartment. Cruciate-retaining total knee arthroplasty (TKA) or posterior cruciate ligament (PCL) reconstruction surgery should aim to achieve stability on the medial side and a few millimeters of laxity at the lateral side at 90 degrees flexion with increasing laxity only on the lateral side in deep flexion.