The in-vivo study of bone graft incorporation has traditionally used a segmental diaphyseal bone defect. This model reliably produces a nonunion, but is complicated by graft instability and altered limb loading stresses. The authors discuss the advantages of a defect-in-continuity canine femur model which produces a more consistent union with fewer mechanical complications despite the absence of fixation. This proposed model permits analysis of radiographic, histologic and biomechanical data which are more applicable to the usual clinical setting in which bone graft is required.