A new technique was used to measure the resultant forces in the anterior cruciate ligament during a series of loading experiments on seventeen fresh-frozen cadaver specimens. The base of the ligament's tibial attachment was mechanically isolated with a coring cutter, and a specially designed load-transducer was fixed to the bone-plug that contained the ligament's tibial insertion so that the resultant forces were directly measured by the load-cell. Although the magnitudes of values for forces varied considerably between specimens for a given test condition, the patterns loading with respect to direction of loading and the angle of flexion of the knee were remarkably consistent. Passive extension of the knee generated forces in the ligament only during the last 10 degrees of extension; at 5 degrees of hyperextension, the forces ranged from fifty to 240 newtons (mean, 118 newtons). When a 200-newton pull of the quadriceps tendon was applied to extend a knee slowly against tibial resistance, however, the force in the ligament increased at all angles of flexion of the knee. Internal tibial torque always generated greater forces in the ligament than did external tibial torque; higher forces were recorded as the knee was extended. The greatest forces (133 to 370 newtons) were generated when ten newton-meters of internal tibial torque was applied to a hyperextended knee. Fifteen newton-meters of applied varus moment generated forces of ninety-four to 177 newtons at full extension; fifteen newton-meters of applied valgus moment generated a mean force of fifty-six newtons, which remained unchanged with flexion of the knee. The force during straight anterior translation of the tibia was approximately equal to the anterior force applied to the tibia. The application of 925 newtons of tibiofemoral contact force reduced the mean force in the ligament that was generated by 200 newtons of anterior pull on the tibia by 36 per cent at full extension and 46 per cent at 20 degrees of flexion. Clinical relevance: Our results have the most direct clinical application when a damaged ligament has been repaired by sutures or augmented with a synthetic stent or autogenous tissue. These data also reflect expected trends of generation of forces in an anterior cruciate-ligament substitute that has been precisely placed at anatomical sites of insertion. The absence of forces in the ligament beyond 10 degrees of flexion of the knee suggests that passive flexion-extension motions in the range from 10 degrees to full flexion would be safe for rehabilitation of the knee after repair or reconstruction of the anterior cruciate ligament. The increases in forces in the anterior cruciate ligament due to pull of the quadriceps tendon indicate that active extension of the quadriceps against resistance in the range of flexion from 0 to 45 degrees would not be advisable if one wanted to limit forces in the anterior cruciate ligament or a ligament substitute. For applied varus moment, the anterior cruciate ligament is markedly more susceptible to elevated forces when the knee is near full extension. Our data indicate that a flexed knee should be less vulnerable to injury to the anterior cruciate ligament caused by applied tibial torque. Joint load acts to protect the anterior cruciate ligament from high forces that are generated by applied straight anterior tibial force; no such protective mechanism was demonstrated for applied internal or external tibial torque.