In this study we investigated the bone-forming capacity of tissue-engineered (TE) constructs implanted ectopically in goats. As cell survival is questionable in large animal models, we investigated the significance of vitality, and thus whether living cells instead of only the potentially osteoinductive extracellular matrix are required to achieve bone formation. Vital TE constructs of porous hydroxyapatite (HA) covered with differentiated bone marrow stromal cells (BMSCs) within an extracellular matrix (ECM) were compared with identical constructs that were devitalized before implantation. The devitalized implants did contain the potentially osteoinductive ECM. Furthermore, we evaluated HA impregnated with fresh bone marrow and HA only. Two different types of HA granules with a volume of approximately 40 microm were investigated: HA70/800, a microporous HA with 70% interconnected macroporosity and an average pore size of 800 microm, and HA60/400, a smooth HA with 60% interconnected macropores and an average size of 400 microm. Two granules of each type were combined and then treated as a single unit for cell seeding, implantation, and histology. The tissue-engineered samples were obtained by seeding culture-expanded goat BMSCs on the HA and subsequently culturing these constructs for 6 days to allow cell differentiation and ECM formation. To devitalize, TE constructs were frozen in liquid nitrogen according to a validated protocol. Fresh bone marrow impregnation was performed perioperatively (4 mL per implant unit). All study groups were implanted in bilateral paraspinal muscles. Fluorochromes were administered at three time points to monitor bone mineralization. After 12 weeks the units were explanted and analyzed by histology of nondecalcified sections. Bone formation was present in all vital tissue-engineered implants. None of the other groups showed any bone formation. Histomorphometry indicated that microporous HA70/800 yielded more bone than did HA60/400. Within the newly formed bone, the fluorescent labels showed that mineralization had occurred before 5 weeks of implantation and was directed from the HA surface toward the center of the pores. In conclusion, tissue-engineered bone formation in goats can be achieved only with viable constructs of an appropriate scaffold and sufficient BMSCs.