Objective To compare the effect of mosaicplasty, mosaicplasty with gene enhanced tissue engineering and mosaicplasty with the gels of non-gene transduced BMSCs in alginate on the treatment of acute osteochondral defects. Methods Western blot test was conducted to detect the expression of hTGF-β1, Col II and Aggrecan in 3 groups, namely hTGF-β1 transduction group, Adv-βgal transduction group and blank control group without transduction. Eighteen 6-month-old Shanghai mascul ine goats weighing 22-25 kg were randomized into groups A, B and C (n=6). BMSCs were isolatedfrom the autologous bone marrow of groups B and C, and were subcultured to get the cells at passage 3. In group B, the BMSCs were transduced with hTGF-β1. For the animals of 3 groups, acute cyl indrical defects 5 mm in diameter and 3 mm in depth were created in the weight bearing area of the medial femoral condyle of hind l imbs. In group A, the autologous osteochondral mosaicplasty was performed to repair the defect; in group B, besides the mosaicplasty, the dead space between the cyl indrical grafts and the host cartilage were injected with the suspension of hTGF-β1 gene transduced autogenous BMSCs in sodium alginate, and CaCl2 was dropped into it to form calcium alginate gels; in group C, the method was the same as the group B, but the BMSCs were not transduced. General condition of the goats after operation was observed, the goats were killed 12 and 24 weeks after operation to receive gross and histology observation, which was evaluated by the histological grading scale of O’Driscoll, Keeley and Salter. Immunohistochemistry and TEM observation were performed 24 weeks after operation. Results Western blot test showed the expression of the hTGF-β1, Col II and the Aggrecan in the hTGF-β1 transduction group were significantly higher than that of the Adv-βgal transduction and the blank control groups. All the goats survived until the end of experiment and all the wounds healed by first intention. Gross observation revealed the boundaries of the reparative tissue in group B were indistinct, with smooth and continuous surfaces of the whole repaired area; while there were gaps in the cartilage spaces of groups A and C. Histology observation showed the dead space between the cyl indrical grafts in group A had fibrocartilage-l ike repair tissue, fill ing of fibrous tissue or overgrowth of the adjacent cartilage; the chondrocytes in group B had regular arrangements, with favorable integrations; while the dead space between the cyl indrical grafts in group C had fibrocartilage-l ike repair tissue, with the existence of gaps. The histology scores of group B at different time points were significantly higher than that of groups A and C, and group C was better than group A (P lt; 0.05); for group B, significant difference was detected between 12 weeks and 24 weeks in the histology score (P lt; 0.05）. Immunohistochemistry staining for Col II 24 weeks after operation showed the chondrocytes and lacuna of the reparative tissue in group B was obviously stained, while groups A and C presented l ight staining. TEM observation showed there were typical chondrocytes in the reparative tissue in group B, while parallel or interlaced arrangement collagen fiber existed in groups A and C. Conclusion Combining mosaicplasty with tissue engineering methods can solve theproblem caused by single use of mosaicplasty, including the poor concrescence of the remnant defect and poor integration with host cartilages.