The treatment of bony defects of the tibia at the time of revision total knee replacement is controversial. The place of compacted morsellised bone graft is becoming established, particularly in contained defects. It has previously been shown that the initial stability of impaction-grafted trays in the contained defects is equivalent to that of an uncemented primary knee replacement. However, there is little biomechanical evidence on which to base a decision in the treatment of uncontained defects. We undertook a laboratory-based biomechanical study comparing three methods of graft containment in segmental medial tibial defects and compared them with the use of a modular metal augment to bypass the defect. Using resin models of the proximal tibia with medial defects representing either 46% or 65% of the medial cortical rim, repair of the defect was accomplished using mesh, cement or a novel bag technique, after which impaction bone grafting was used to fill the contained defects and a tibial component was cemented in place. As a control, a cemented tibial component with modular metal augments was used in identical defects. All specimens were submitted to cyclical mechanical loading, during which cyclical and permanent tray displacement were determined. The results showed satisfactory stability with all the techniques except the bone bag method. Using metal augments gave the highest initial stability, but obviously lacked any potential for bone restoration