This study tested whether different in vitro cultivation techniques for tissue-engineered scaffolds seeded with human trabecular bone cells affect in vivo bone formation when implanted into critical-size defects in rat mandibles. Human trabecular cells were isolated and seeded into three types of scaffolds (porous CaCO(3), mineralized collagen, porous tricalcium phosphate). Four in vitro groups were produced: empty control scaffolds incubated with cell culture medium for 24 h; scaffolds seeded with trabecular bone cells, cultivated under static conditions for 24 h; scaffolds seeded with trabecular bone cells, cultivated for 14 days under static conditions; scaffolds seeded with trabecular bone cells, cultivated for 14 days in a continuous flow perfusion bioreactor. The scaffolds were implanted press fit into non-healing defects, 5 mm diameter, in rat mandibles. After 6 weeks the presence of human cells was assessed; none were detected. Histomorphometric evaluation showed that neither seeding human trabecular bone cells nor the culturing technique increased the amount of early bone formation compared with the level provided by osteoconductive bone ingrowth from the defect edges. It is concluded that human bone marrow stroma cells in tissue-engineered scaffolds and associated in vitro technology are difficult to test in the mandible in animal models.