The strain developed in the posterior cruciate ligament (PCL) of eight fresh cadaveric knees was measured before and after total knee arthroplasty using a loading technique that simulated stair ascent and descent. Each knee was instrumented with a Hall Effect strain gauge (Micro-Strain, Burlington, VT) in the PCL, a load cell in the quadriceps tendon, an electrogoniometer, and an array of linear displacement transducers to measure femoral rollback. Testing was undertaken with each knee in its normal state with the anterior cruciate cut and with a cruciate-retaining prosthesis, a cruciate-excising prosthesis, and a cruciate-substituting prosthesis. Normal PCL strain levels were produced in only 37% of the trials following implantation of the cruciate-retaining knee arthroplasties. With a cruciate-retaining prosthesis, femoral rollback decreased by an average of 36% and was associated with a 15% loss in extensor efficiency. In the procedures performed with excision of the PCL, rollback decreased by 70% and extensor efficiency by 19%. Cruciate substitution resulted in a 12% loss in rollback and an 11% decrease in extensor efficiency. The strain developed within the PCL during knee flexion was found to be extremely sensitive to the thickness of the polymeric tibial insert. In the majority of cases, it was not possible to restore normal ligament loading with flexion while simultaneously maintaining acceptable varus/valgus stability of the knee joint. Using a range of contemporary knee arthroplasties, the authors were unable to consistently reproduce normal function of the PCL.