Interbody fusion is increasingly widely used as a treatment for intervertebral disc disorders, but the biomechanics of the procedure are not well understood. The compressive loads through the spine are largely carried by the implant or bone graft, which typically rests on a relatively small area of the vertebral body. As the compressive strength of the bone is very low, subsidence of the implants into the vertebral bodies is a common clinical complication. Previous biomechanical studies of spinal fusion have concentrated on the stiffness of the constructs, which is important in promoting fusion. Preliminary studies have shown that there are large differences in compressive strength between different implant systems, and gave an insight into the biomechanical factors that are important in determining the strength of spinal fusion constructs. This paper reports part of a larger on going study comparing anterior and posterior fusion systems, with various methods of fixation. A major problem in interpreting the results of these tests is to distinguish between initial settling of the implants and the onset of failure to construct. We have developed a novel technique using acoustic emission monitoring to detect microcracking in the bones, which allows the onset of failure to be distinguished from initial bedding in of the implants. Two implant systems were tested, the Syncage and the Contact fusion cage. The cages were implanted into porcine lumbar spines at L4-L5, and the implanted motion segment was then dissected out. Steel plates were mounted on each end using bone cement to ensure an even distribution of load through the vertebral body. The complete constructs were then loaded in compression, using acoustic emission sensors to detect microcracking in the bones. The load was cyclically increased in o.5kN steps until failure occurred. The acoustic emission technique gave a sensitive indication of the onset of damage in the bones and allowed the initial settling of the implant under load to be identified. Using cyclic unloading and reloading, it was possible to accurately identify whether this damage had weakened the construct or increased its strength by redistributing stress concentrations. Initial results indicate that the Contact fusion cage fails at a much lower load than the Syncage in this model; this is ascribed to the very small contact areas between the cage and the vertebral body, which results in high compressive stresses in the bone. Under large compressive loads it appears that the constructs become unstable, and fail by buckling and plastic collapse of the vertebral bodies. Various failure models are therefore possible depending on which part of the vertebral body starts to collapse first.