Abstract : Large bone defects are still a challenge to orthopedic surgeons. In this study, a massive bone defect with a clinically relevant volume was efficiently reconstructed by transplanting an engineered bone in which mesenchymal stem cells (MSCs) expanded in autologous serum (AS) were combined with a porous scaffold. In the first step, we established that the way in which the MSCs are distributed over the scaffold affects the ultimate bone-forming ability of the transplant: constructs consisting of a natural coral scaffold and a pseudo-periosteal layer of MSCs surrounding the implant (coral-MSC3D) formed significantly more bone than constructs in which the MSCs were distributed throughout the implant (p = 0.01). However, bone healing occurred in only one sheep, owing to the high resorption rate of natural coral scaffold. To overcome this problem, constructs in which MSCs were combined with a porous coralline-based hydroxyapatite (CHA) scaffold having the same architecture as natural coral but a lower resorption rate were prepared. After their implantation, these constructs were found to have the same osteogenic potential as autologous bone grafts in terms of the amount of newly formed bone present at 4 months (p = 0.89) and to have been completely replaced by newly formed, structurally competent bone within 14 months. Nevertheless, although the rate of bone healing was strikingly improved when CHA-MSC3D constructs were used (five of seven animals healed) as compared with the coral-MSC3D construct (one of seven healed), it was still less satisfactory than that obtained with autografts (five of five healed).