OBJECTIVE Novel approaches to intervention in joint diseases consist of the replacement of diseased cartilage by in vitro engineered, viable cells or graft tissues. Two major obstacles remain to be overcome: (1) Hyaline cartilage in vitro often loses differentiated traits. (2) Grafts frequently are not integrated satisfactorily into host cartilage and/or the tissue is remodelled in situ into functionally inferior fibrocartilage. Therefore, we have explored the possibility whether chondrocytes embedded into agarose gels provided better graft tissues in a repair model of full thickness defects in rabbit joint cartilage. METHODS Experimental defects of knee joint cartilage was filled with articular chondrocytes cultured in agarose gels. Chondrocytes in vitro either remained unstimulated or were treated with several growth factors. Repair of the defects was assessed by histology and was scored between 0 (no healing) and 1 (perfect healing) as judged by the follwing parameters: intensity of proteoglycan staining, organization of the superficial zone, ossification at the border between repair cartilage and subchondral bone, tidemark formation in the repaired area, arrangement of chondrocytes, and integration of repair cartilage into host. RESULTS Treatment of chondrocyte cultures with bFGF had a stabilizing effect on the differentiated state of the cells in implanted grafts whereas bone morphogenetic proteins stimulated ingrowth of subchondral bone reducing repair cartilage thickness and preventing normal tide mark formation; TGF-beta did not significantly affect evaluation parameters in comparison with untreated controls. CONCLUSIONS Growth factor treatment resulted in an ambiguous quality of graft development. Only FGF had a clear beneficial effect to the graft tissues after 1 month. Further studies are required to define the precise conditions and sequence of growth factor treatment of in vitro engineered cartilage which benefits graft quality.