Active and passive components of torsional stability of the knee were measured with an instrumented clinical knee-testing apparatus. Torque-versus-rotation response curves were recorded in the non-weight-bearing condition with muscles relaxed for twenty normal subjects who were tested at 20 and 90 degrees of knee flexion with the hips flexed and extended. At applied torque levels as high as +/-10 newton-meters, tibial rotation averaged approximately one-half the foot rotation. The mean algebraic right-left rotation difference for the group was nearly zero; however, sizable standard deviations for this difference indicated considerable right-left variations between individuals in the test group. Maximum isometrically generated tibial torques were measured by asking the subjects to twist with an explosive effort against a locked torque-cell. No significant differences in generated torque were measured between preferred and non-preferred lower limbs, with only one minor exception. Subjects generally were able to generate greater internal torque than external torque. When the foot was locked in a position of internal or external rotation, an individual was able to generate increased tibial torque in the direction that would tend to return the foot to the neutral position. Flexion of the knee from 20 to 90 degrees increased externally generated torque, while internal torque was affected to a lesser degree. Flexion of the hip had little effect on generated torque. Six cadaver knees without menisci that were tested to failure in external rotation showed torque levels for ligament failure to be similar in magnitude to the maximum generated isometric torque that acts to protect the knee ligaments.