The use of a bovine bone-derived osteoinductive protein extract as a bone-graft substitute was evaluated in a rabbit model of intertransverse process arthrodesis of the lumbar spine. Forty-five adult New Zealand White rabbits had arthrodesis between the fifth and sixth lumbar vertebrae with use of one of three graft materials: autogenous iliac-crest bone, osteoinductive protein delivered in an allogeneic demineralized bone matrix/collagen carrier, or demineralized bone matrix/collagen carrier or demineralized bone matrix/collagen carrier without osteoinductive protein. Fusion was assessed by manual palpation, radiography, biomechanical testing, and light microscopy at two and five weeks after the operation. At two weeks, light microscopic analysis of the arthrodesis site in which osteoinductive protein had been used showed that most of the demineralized bone matrix was still present, with small amounts of membranous and endochondral bone formation at the peripheral margins of the implant. Light microscopic analysis of the five-week specimens showed increased new-bone formation and a more homogeneous and mature fusion mass with the osteoinductive bone protein than with the autogenous bone graft. At five weeks, the fusions with the osteoinductive protein extract were characterized by more secondary spongiosa, with formation of bone marrow centrally and a cortical rim peripherally. Of the thirty-five rabbits that were examined at five weeks, all ten in the group that had received osteoinductive bone protein had a solid fusion, but the rate of fusion was significantly less in the other two groups: eight of thirteen rabbits (p = 0.05) in the group that had received autogenous bone graft and two of twelve rabbits (p = 0.0001) in the group that had received demineralized bone matrix/collagen carrier without osteoinductive bone protein. The use of osteoinductive bone protein resulted in stronger (p = 0.02) and stiffer (p = 0.005) fusions compared with those obtained with the use of autogenous iliac-crest graft.