Ten knees were studied using a robotic testing system under a 134-N posterior tibial load at five flexion angles. Three knee positions were used to study the effect of flexion angle at the time of graft fixation (full extension, 60 degrees, and 90 degrees) and two were used to study the effect of anterior tibial load (60 degrees and 90 degrees). Knee kinematics and in situ forces were determined for the intact ligament and the graft for each reconstruction. Graft fixation at full extension significantly decreased posterior tibial translation compared with the intact knee by up to 2.9 +/- 2.9 mm at 30 degrees, while in situ forces in the graft were up to 18 +/- 35 N greater than for the intact ligament. Conversely, posterior tibial translation for graft fixation at 90 degrees was significantly greater than that of the intact knee by up to 2.2 +/- 1.1 mm at all flexion angles; in situ forces decreased as much as 33 +/- 30 N. When an anterior tibial load was applied before graft fixation at 90 degrees of flexion, posterior tibial translation did not differ from the intact knee from 30 degrees to 120 degrees, while the in situ force in the graft did not differ from the intact ligament at full extension, 60 degrees, and 120 degrees of flexion. These data suggest that graft fixation at full extension may overconstrain the knee and elevate in situ graft forces. Conversely, fixation with the knee in flexion and an anterior tibial load best restored intact knee biomechanics.