Bone tissue engineering is currently undergoing a paradigm shift regarding the concepts used to develop cell-based therapies for skeletal repair. In place of the "trial and error" approach, researchers aim at developing cellular concepts that mirror developmental and postnatal processes. Herein, we describe a model for in vivo endochondral remodeling of an in vitro derived cartilaginous intermediate and its applicability to bone engineering. In vitro differentiation of the continuous cell line, ATDC5, in pellet culture was enhanced in a medium containing ascorbic acid, insulin-transferrin-selenium, dexamethasone, and transforming growth factor β1, when compared with other tested preparations. This differentiation was characterized by the elevated expression of Collagen type II and X along with glycosaminoglycan (GAG) accumulation and the onset of hypertrophy. On combination with NuOss™, a clinically used bone void filler, and implantation in nude mice, the differentiated pellets further matured into GAG rich cartilaginous intermediates after 4 weeks. This was subsequently partially remodeled into osteocalcin-positive bone tissue after 8 weeks without further external manipulation, indicating the semi-autonomous nature of this implant. Mineralized tissue along with active osteoclast resorption and neo-angiogenesis was apparent throughout the implant. The bone volume was approximately eightfold higher (10.70%±0.99%) when using a cartilaginous intermediate (based on differentiated cell pellets) than when observed with cell-seeded scaffolds (1.19%±0.24% and 1.48%±0.35%), in both a differentiated and an undifferentiated state. This study highlights the potential of endochondral strategies for bone tissue engineering and allows the identification of the key cellular parameters for this process.