This study was undertaken to determine the potential of demineralized bone alloimplant (DBA) for healing long bone defects. We removed 25 mm of bone from the junction of the proximal and middle thirds of both ulnas in seven adult male mongrel dogs. On one side of each animal the devascularized bone was replaced as a cortical autograft (CA) and on the other side, the defect was filled with DBA. Because of implant error, there were only five CAs to evaluate, of which two had achieved clinical union. Of the seven DBAs only one achieved union. The bone in and around the defect was quantified on each specimen roentgenogram and was compared to the size of the original defect. The CA defects were filled 139% +/- 11% with bone that was partly new and partly remnants of the replaced autograft. The demineralized alloimplant defects were filled 143% +/- 45% with bone which, of course, was all newly formed. The difference between the two groups was not significant. We concluded that the DBA was an excellent osteogenic stimulant, but that it did not prevent nonunion in this model.