The purpose of the present study was to test the hypothesis that cortical bone loss, trabecular bone density and the amount of bone ingrowth vary as a function of stem stiffness in a canine cementless hip replacement model. The study was motivated by the problem of cortical bone atrophy in the proximal femur following cementless total hip replacement. Two stem stiffnesses were used and both designs were identical in external geometry and porous coating placement. The high stiffness stem caused approximately 26% cortical bone stress-shielding and the low stiffness stem caused approximately 7.5% stress-shielding, as assessed by beam theory. Each group included nine adult, male canines who received unilateral arthroplasties for a period of six months. The animals with the low stiffness stems tended to lose less proximal cortical bone than the animals with high stiffness stems (4% +/- 9 as opposed to 11% +/- 14), but the difference was not statistically significant (p = 0.251). However, the patterns of bone ingrowth into the implant and change in medullary bone density adjacent to the implant were fundamentally different as a function of stem stiffness (p < 0.01). Most importantly, while the high stiffness group had peaks in these variables at the distal end of the stem, the low stiffness group had peak values proximally. These different patterns of functional adaptation are consistent with the idea that reduced stem stiffness enhances proximal load transfer.