Objectives. Orthopaedic surgeons use stems in revision knee surgery to obtain stability when metaphyseal bone is missing. No consensus exists regarding stem size or method of fixation. This in vitro study investigated the influence of stem length and method of fixation on the pattern and level of relative motion at the bone–implant interface at a range of functional flexion angles. Methods. A custom test rig using differential variable reluctance transducers (DVRTs) was developed to record all translational and rotational motions at the bone–implant interface. Composite femurs were used. These were secured to permit variation in flexion angle from 0° to 90°. Cyclic loads were applied through a tibial component based on three peaks corresponding to 0°, 10° and 20° flexion from a normal walking cycle. Three different femoral components were investigated in this study for cementless and cemented interface conditions. Results. Relative motions were found to increase with flexion angle. Stemmed implants reduced relative motions in comparison to stemless implants for uncemented constructs. Relative motions for cemented implants were reduced to one-third of their equivalent uncemented constructs. Conclusions. Stems are not necessary for cemented implants when the metaphyseal bone is intact. Short cemented femoral stems confer as much stability as long uncemented stems