Background. The gold standard of bone graft substitutes is vascularized autologous bone. Our strategy to acquire this bone substitute is using recombinant human bone morphogenetic protein (rhBMP)-2 with beta-tricalcium phosphate (TCP) as carrier to induce bone tissue in muscle and grafting as a muscle-pedicled bone flap. In our previous studies, morphologically controlled living bone was successfully induced in a rat muscular pedicle. Moreover, concurrent administration of bisphosphonate controlled the subsequent resorption of the induced bone. This study was conducted (1) to confirm that when the size of carrier is increased, the induced bone can form the same in the center as well as on the periphery of the carrier, and (2) to simulate in vivo banking for vascularized autograft bone. Methods. In experiment 1200 mu g rhBMP-2 and 10(-5)M bisphosphonate (Minodronate) were applied to a 1-cm-diameter sphere of beta-TCP. This sphere was then inoculated into rabbit quadriceps and harvested 5 weeks later (n = 3). In experiment 2, beta-TCP was generated as a one-third hollow cylinder with 4-mm outer radius, 2-mm inner radius, and 10-mm length. After 50 mu g rhBMP-2 and 10(-5)M Minodronate were applied, this structure was then inoculated into rat quadriceps. Four weeks later, the induced bone was transplanted as a muscle-pedicled flap onto the adjacent femur and harvested after 4 and 8 weeks more (n = 3). Results. In experiment 1, good bone formation was observed at the central as well as the peripheral region of the sphere. In experiment 2, good bone formation and bony union to the femur at both 4 and 8 weeks were detected on the X-ray film. The presence of new bone formation around the screw was observed. Conclusions. Transplantation of the beta-TCP carrier treated with rhBMP-2 and bisphosphonate into the muscular tissue may have clinical potential for in vivo banking for muscle-pedicled bone flaps.