Aims: To investigate the optimal dimension interference screw for fixation of a tendon graft in a bone tunnel. Methods: A porcine model was developed to represent ACL reconstruction in the distal femur. A standard 8mm size pig flexor tendon graft was inserted into a standard 8mm bone tunnel. The screw dimensions were varied. The tendon was loaded with a constant force using a Universal Testing Machine. The construct was tested to failure at a rate of 50mm/minute. Load, deformation data and mechanism of construct failure were recorded. The screw diameters of 7, 8 and 9mm and lengths of 20, 25 and 30mm were tested in 80 individual reconstructions. Results: The mean pull out force was similar between the 7mm (191N) and 8mm screws (188N), but significantly different for the 9mm screw (109N) (p< 0.05) The 30mm screw (231N) was marginally better than the 20mm screw (215N) (p> 0.05). The mechanism of failure however, was significantly different between the groups. All grafts fixed with a 9mm screw failed at the tunnel opening (100%), whereas those fixed with a 7mm screw failed by slippage of the graft along the length of the tunnel (83%). The screw length did not affect the mechanism of graft failure. Conclusion: Our results suggest that a screw size equal to or 1mm less than the diameter of the tunnel gives the optimum initial interference screw fixation of tendon in a bone tunnel.