A variety of designs of interlocking intramedullary rods has been developed in recent years for fixation of otherwise unstable femoral fractures. In this study, plastic model femurs with either stable transverse subtrochanteric osteotomies or unstable subtrochanteric defects were used to compare the rotational stiffness provided by different types of intramedullary rods. Simultaneous axial, bending, and torsional loading was applied to examine the effect of rod-bone and fracture surface interaction on rotational stiffness. For the stable transverse fracture, increasing the axial load from 150 to 1000 N increased the rotational stiffness as much as 5.7 times (Zickel nail). Thus, a large component of the resistance to rotation was due to interaction of the fracture surfaces. In contrast, with the unstable segmental defect, rotational stiffness was primarily dependent on the inherent torsional stiffness of the implant. For example, the fracture fixed with a double-interlocked tubular titanium rod was about seven times stiffer in rotation than fractures fixed with slotted-cloverleaf intramedullary rods. Close comparison of these experimentally measured characteristics with actual clinical performance will be necessary to identify the optimum design features for interlocking intramedullary devices.