OBJECTIVES: To compare the mechanical properties of locking compression plate (LCP) and limited contact dynamic compression plate (LC-DCP) constructs in an experimental model of comminuted fracture of the canine femur during eccentric cyclic loading. METHODS: A 20 mm mid-diaphyseal gap was created in eighteen canine femora. A 10-hole, 2.4 mm stainless steel plate (LCP or LC-DCP) was applied with three bicortical screws in each bone fragment. Eccentric cyclic loadings were applied at 10 Hertz for 610,000 cycles. Quasistatic loading / unloading cycles were applied at 0 and 10,000 cycles, and then every 50,000 cycles. Structural stiffness was calculated as the slope of the linear portion of the load-displacement curves during quasistatic loading / unloading cycles. RESULTS: No bone failure or screw loosening occurred. Two of the nine LCP constructs failed by plate breakage during fatigue testing, whereas no gross failure occurred with the LC-DCP constructs. The mean first stiffness of the LCP constructs over the course of testing was 24.0% lower than that of constructs stabilized by LC-DCP. Construct stiffness increased in some specimens during testing, presumably due to changes in bone-plate contact. The first stiffness of LC-DCP constructs decreased by 19.4% and that of locked constructs by 34.3% during the cycling period. A biphasic stiffness profile was observed: the second stiffness was significantly greater than the first stiffness in both groups, which allowed progressive stabilization at elevated load levels. CLINICAL SIGNIFICANCE: Because LCP are not compressed to the bone, they may have a longer working length across a fracture, and thus be less stiff. However, this may cause them to be more susceptible to fatigue failure if healing is delayed.