Recombinant human bone morphogenetic proteins (rhBMPs) are often used during spine surgery, but their effects on postoperative infections have not been well elucidated. Long-bone studies suggest that BMPs may limit local infection and facilitate bone formation. Until now, rhBMP-2 had not been evaluated in the setting of infected spinal arthrodesis. In the study reported here, we evaluated the safety and efficacy of rhBMP-2 and autograft in inducing fusion in the setting of surgically acquired infection. Sixty rabbits underwent fusion with autograft or rhBMP-2 with coadministration of Staphylococcus aureus or sterile saline. In the noninoculated groups, 4/15 autograft and 13/13 rhBMP-2 rabbits fused (P<.001). In the inoculated groups, 0/14 autograft and 3/12 rhBMP-2 rabbits fused (P = .085). There were 4/14 early deaths caused by infection in the autograft group and 0/12 in the rhBMP-2 group (P = .1). Although the difference in fusion rates and early deaths from infection for rhBMP-2 and autograft did not reach our predetermined alpha error threshold, the data were trending toward significance. Our results demonstrated no increase in morbidity or mortality associated with use of rhBMP-2 in the setting of local infection. Although BMP use with infections remains controversial, these results indicate that rhBMP- 2 could be used in a contaminated environment.