We developed an experimental model to compare the efficacy of free vascularized bone grafts, conventional segmental autografts, matchstick autografts, and fresh segmental allografts in terms of their ability to reconstruct a 7-cm segmental diaphyseal defect created in the canine femur. Forty-five adult mongrel dogs were studied and followed for 6 to 12 months prior to sacrifice. Evaluation included radiologic assessment of graft incorporation and hypertrophy, histology, and biomechanical testing. These studies indicated that microsurgically revascularized autografts were superior to all other groups in terms of early incorporation, hypertrophy, and the highest mechanical strength to failure. Union of the bone graft to the recipient femur was achieved by 6 months in 25 of 26 autografts, and no difference in union rate was seen within the autograft group. However, only two of five allografts achieved bony union during this time interval. Arteriography, microangiography, fluorochrome, and histologic studies all supported the concept that microsurgically revascularized grafts, when successful, maintain their viability. However, the premise that all osteocytes survive in a successfully revascularized bone graft is open to question. While decalcified sections showed that all microsurgically revascularized grafts maintained normal viability in the central marrow and cancellous portions compared with the other three groups, the viability of cortical bone in the vascularized autografts was less clear. Undecalcified fluorochrome sections suggested that circulation was not preserved in all portions of the cortex. Revascularization of the nonvascularized autografts was complete at 3 months, while, in the avascular allografts, the process was not complete at 6 months.