The effects of several parameters on the accuracy and reproducibility of anterior-posterior (AP) drawer measurements were investigated. An instrumented knee-drawer tester for AP laxity evaluations was constructed, based on the differential displacement method, measuring shifts of the tuberosity relative to the patella. The accuracy of the AP-shift was determined with the parallel use of a highly accurate roentgen stereo photogrammetric (RSP) measurement system on two postmortem leg specimens. The effects of relative motion between patella and femur were negligible. In addition to AP shifts, significant knee flexion and tibial rotations occurred. Tests were made to evaluate the effects of different observers, time sequences, and different days. In addition, effects of muscle relaxation were studied. Overall, the shift parameters at different forces were found to be reproducible to between 5 and 15%. The slopes (compliances) of the laxity curves, at different forces, were found to be reproducible to between 20 and 40%. The reproducibility was principally affected by deviations in the subject positioning procedure.