The increasing clinical use of human demineralized bone matrix has brought about the desire to understand better the osteoinductivity of these graft materials. The rat heterotopic model has been used successfully to show the sequence of events involved in the endochondral ossification process resulting in osteoinduction. In this study, the osteoinductive potential of human demineralized bone powder was assessed, using immune compromised rats (athymic rnu/rnu) to avoid problems associated with cross species incompatibilities. Implants were placed in subcutaneous or intermuscular sites. This model is characterized to provide a basis for routinely determining the performance of human demineralized bone powder. Demineralized bone powder was prepared from rat and human cortical bone according to a strict protocol. The lack of response to guanidine HC1 extracted (noninductive) demineralized bone showed the selectivity of the assay. The same lots of human and rat demineralized bone were tested in sequential experiments during a 1-year period. These results showed reproducible induction of the demineralized bone powder between experiments. Combining demineralized bone with the guanidine HC1 extracted demineralized bone in varying ratios tested the sensitivity of the assay. These results showed an increase in bone formation with increasing quantities of active demineralized bone and established the ability of the bioassay to differentiate between the various levels of active (osteoinductive) demineralized bone powder. With this model, consistent performance of demineralized bone powder processed by well controlled methods was seen.