In vivo bioreactors serve as regenerative niches that improve vascularization and regeneration of bone grafts. This study has evaluated the masseter muscle as a natural bioreactor for βTCP or PCL/βTCP scaffolds, in terms of bone regeneration. The effect of pedicle preservation, along with sole, or MSC‐ or rhBMP2‐combined application of scaffolds, has also been studied. Twenty‐four mongrel dogs were randomly placed in six groups, including βTCP, βTCP/rhBMP2, βTCP/MSCs, PCL/βTCP, PCL/βTCP/rhBMP2, and PCL/βTCP/MSCs. During the first surgery, the scaffolds were implanted into the masseter muscle for being prefabricated. After 2 months, each group was divided into two subgroups prior to mandibular bone defect reconstruction; one with a preserved vascularized pedicle and one without. After 12 weeks, animals were euthanized, and new bone formation was evaluated using histological analysis. Histological analysis showed that all β‐TCP scaffold groups had resulted in significantly greater rates of new bone formation, either with a pedicle surgical approach or non‐pedicle surgical approach, comparing to their parallel groups of βTCP/PCL scaffolds (p ≤ .05). Pedicled β‐TCP scaffold groups that were treated with either rhBMP2 (48.443% ± 0.250%) or MSCs (46.577% ± 0.601%) demonstrated the highest rates of new bone formation (p ≤ .05). Therefore, masseter muscle can be used as a local in vivo bioreactor with potential clinical advantages in reconstruction of human mandibular defects. In addition, scaffold composition, pedicle preservation, and treatment with MSCs or rhBMP2, influence new bone formation and scaffold degradation rates in the prefabrication technique.