An experimental model of a contaminated open fracture has been developed. This model has been used to test the hypothesis that stable fixation of a contaminated open fracture will reduce its susceptibility to infection. The tibiae of male New Zealand white rabbits were fractured and then fixed with either a dynamic compression plate (stable group) or a loose-fitting intramedullary rod (unstable group). The fracture site was then inoculated with a standard inoculum of Staphylococcus aureus. There were 20 rabbits in the stable group and osteomyelitis developed in seven (35 per cent); in the unstable group 15 (71 per cent) out of 21 animals developed osteomyelitis. This difference in infection rates was statistically significant (P < 0.02). This experimental study supports the concept of stabilization of open fractures in man.