Torsional loads of daily activities contribute to the failure of the primary fixation of hip prostheses. Implant torsional stability must be evaluated prior to in vivo clinical trials. Whereas previous work has investigated this phenomenon, descriptions of physiologically accurate and reproducible in vitro methodologies are rare. The present study aimed to detect and control the sources of error and variability that influence in vitro methods. A typical set-up for the analysis of primary stability of hip stems was studied. Load cycles included proximal-to-distal axial force, torque, and bending moment. The effects of loading frequency and strain distribution across the cortical bone were investigated in order to optimize testing conditions and measurement set-up. The relative shear motion at the bone-stem interface was measured transcortically using linear variable displacement transducers. The procedures developed for mounting specimens on the testing machine and for positioning sensors on the specimen were standardized and tested for reproducibility. The protocol was finally tested for repeatability and accuracy. Measurement errors were 2.3 microm between load cycles and 4.9 microm for repeated set-ups, comparing favorably with the literature.