Nonunion treatment has a high rate of success, although recalcitrant nonunion may determine the need for amputation. Therefore, new treatment options are continuously investigated in order to further reduce the risk of nonunion recurrence. This study aimed to (a) develop a new large animal model for bone atrophic nonunion and (b) compare the efficacy of demineralized bone matrix (DBM) and DBM in combination with mesenchymal stem cells (MSC) in the new nonunion model. The new model consists of a noncritical, full-thickness segmental defect created in the sheep tibia, stabilized by an intramedullary nail, and involves the creation of a locally impaired blood supply achieved through periosteum excision and electrocauterization of the stump ends. Six weeks after defect creation, lack of hard tissue callus and established nonunion was observed in all operated tibiae both by radiographic and clinical evaluation. Nonunion was treated with allogeneic DBM or autologous MSC cultivated on DBM particles (DBM + MSC) for 1 day before implantation. Twelve weeks after treatment, radiographic, microtomographic, histologic, and histomorphometric analysis showed the formation of bone callus in DBM group, whereas the fracture healing appeared at an early stage in DBM + MSC group. Torsional strength and stiffness of the DBM group appeared higher than those of DBM + MSC group, although the differences were not statistically significant. In conclusion, a new sheep bone nonunion model resembling the complexity of the clinical condition was developed. DBM is an effective option for nonunion treatment, whereas MSC do not improve the healing process when cultivated on DBM particles before implantation.