Eleven fresh frozen cadaveric knee specimens were mounted in a knee kinematics test device, and normal patellar movements were evaluated with use of an external device for direct measurement of patellar movements. The effects of four different measurement conditions were assessed through alteration of one condition and determination of its effect on patellar kinematics with the use of six specimens. The four conditions included (a) change of the measuring axis from an axis parallel to the central axis of the femur (femoral axis) to one parallel to the central axis of the tibia (tibial axis), (b) rotation of the femoral axis internally 6", (c) change of the direction of the quadriceps force from parallel to the mechanical line of the lower extremity to a direction parallel to the femoral shaft, and (d) increase of the magnitude of the quadriceps force from 111 to 500 N. During knee flexion, the patella shifted laterally after a slight initial medial shift, tilted laterally from midflexion to 90Β°, and gradually rotated medially. The patellar shift relative to the tibial axis appeared to be more medial than the shift measured relative to the femoral axis; the discrepancy was caused by the valgus position of the tibia relative to the femur. Changing the rotational angle of the femoral axis artificially changed the patellar position. Varying the direction of the quadriceps within the narrow range and increasing the quadriceps force did not affect patellar movements. It is likely that the anatomic configuration of the patella allows the patella to seat in a stable configuration so that it resists moderate changes in the load and direction of the quadriceps. Tibia1 rotation exerted a major influence on patellar shift and tilt in the early phase of knee flexion: the patella rotated medially when the tibia was externally rotated and rotated laterally when the tibia was internally rotated. These results indicate that the patella may be unstable in the first phase of knee flexion when the tibia is rotated.