IntroductionPosterior lumbar interbody fusion (PLIF) was first described by Briggs and Milligan in 1944 [4] and modified by Cloward in 1953 [6]. PLIF is a surgically sophisticated technique for arthrodesis and stabilisation within the intradiscal space of an unstable lumbar segment. It consists of a generous posterior decompression of the neural structures and then a discectomy and anterior interbody arthrodesis. The intradiscal work is performed between the exiting nerve root and the retracted transversal nerve root and thecal sac at the lumbar motion segment to be stabilised. Indications for this procedure include degenerative lumbar instability and spondylolisthesis [6,11,12,15]. This surgical approach has the advantage over posterior arthrodesis of obtaining a mature fusion in the anterior column, which bears 80% of the axial load and is under compression in an upright posture [20]. Cloward's technique [6] consisted of interbody fusion using autologous iliac crest bone grafts without additional internal stabilisation. The major disadvantage of this technique is the minimal initial internal stabilisation. Furthermore, the stiffness of these constructs decreases in the first few months during the graft absorption and initial integration phase. Graft-related complications with this technique have been reported to range from 3 to 18% [5,16,22]. To improve these initial results with PLIF operations, a variAbstract A high rate of pseudarthrosis and a high overall rate of implant migration requiring surgical revision has been reported following posterior lumbar interbody fusion using BAK threaded cages. The high rate of both pseudarthrosis and implant migration may be due to poor fixation of the implant. The purpose of this study was to analyse the motion of threaded cages in posterior lumbar interbody fusion. Six cadaveric human lumbar spine segments (three L2/3 and three L4/5 segments) were prepared for biomechanical testing. The segments were tested, without preload, under forces of axial compression (600 N), torsion (25 Nm) and shearing force (250 N).The tests were performed first with the segments in an intact state, and subsequently following instrumented stabilisation with two BAK cages via a posterior approach. These results were compared with those of a finite element model simulating the effects of identical forces on the segments with constructs. As the results were comparable, the finite element model was used for analysing the motion of BAK cages within the disc space. Motion of the implants was not seen in compression. In torsion, a rolling motion was noted, with a range of motion of 10.6°around the central axis of the implant when left/right torsion (25 Nm) was applied. The way the implants move within the segment may be due to their special shape: the thread of the implants can not prevent the BAK cages rolling within the disc space.