Controlled bending moments were applied to twelve human cadaver knee joints using a special loading fixture that allowed variation of both the Q-angle and the flexion angle. The joints were tested at three different Q-angles (physiological, increased 10 degrees, and decreased 10 degrees) and five different angles of flexion (ranging from 20 to 120 degrees). Based on one-third of values in the literature for maximum voluntary isometric quadriceps moments, we applied resultant knee moments of 23.6, 30.7, 47.2, and 35.0 newton-meters at 20, 30, 60, and 90 degrees of flexion, respectively. Based on two-thirds of reported maximum moments, we applied 47.2 newton-meters at 120 degrees of flexion. Normal patellofemoral-contact pressures, measured with a pressure-sensitive film, were remarkably uniformly distributed (+/- 0.25 megapascal ), with approximately the same pressure on the lateral and medial patellar facets. The maximum contact force occurred at 90 degrees of flexion. Extrapolating our measurements to full in vivo moments, we estimated maximum contact forces of 4600 newtons , or approximately 6.5 times body weight. Tendofemoral contact at 120 degrees of knee flexion supported one-third of the total contact force on the patella. A 10-degree increase in the Q-angle resulted in increased peak pressures (an increase of 45 per cent at 20 degrees of flexion). A decrease in the Q-angle resulted in unloading of the vertical crest and, in some knees, of parts of the lateral facet. However, these decreases were always associated with increased peak pressures (50 per cent more at 20 degrees of flexion) at other locations.