Bone–implant displacements can be caused by rigid body motion and by differences in material properties of the implant and bone. In the present study of the tibial component in total knee replacement, we tested a series of tibial component fixation designs to determine how certain design features influenced the magnitude of the tangential displacement between the component and supporting bone in a canine model. The transverse expansion of the proximal tibia under static axial loading was measured in the intact tibia and then in the same bone following implantation of tibial components with different interface characteristics: cementless flat smooth, cementless flat porous-coated, cementless flat porous-coated with screws, cementless pegged porous-coated, cementless pegged porous-coated with screws, cemented pegged, and cemented pegged with screws. In all cases, the magnitude of the transverse expansion increased with higher applied loads. When the statistical analysis was restricted to the cementless interfaces, the presence/absence of the porous coating, the presence/absence of pegs, and the use of screws had no significant influence on tibial expansion. However, in an analysis including the cemented and cementless pegged components, tibial expansion was reduced with the use of screws. The magnitude of the interface motion due to these displacement incompatibilities was approximately fivefold lower than the amount of interface motion related to rigid body motion found in a separate study with the canine model. The measured expansion was similar in the intact tibiae and the implanted tibiae, suggesting that the transverse constraint in the canine proximal tibia must be provided by the surrounding cortical ring rather than the subchondral bone. © 1994 John Wiley & Sons, Inc.