Full-thickness articular cartilage damage does not resolve spontaneously. Studies with growth factors, implantation of autologous chondrocytes and mesenchymal stem cells have led to variable, to some extent inconsistent, results. This work compares osteochondral knee-defect repair in rabbits upon implantation of a previously described alginate/ (poly(lactic-co-glycolic) acid (PLGA) osteochondral scaffold in distinct conditions. Systems were either in vitro pre-cultured with a small number of allogeneic chondrocytes under fibroblast growth factor (FGF)-2 stimulation or the same amount of allogeneic, marrow derived, mesenchymal stem cells (without any pre-differentiation), or loaded with microsphere-encapsulated bone morphogenetic protein (BMP)-2 within the alginate layer, or holding combinations of one or the other cell type with BMP-2. The experimental limit was 12 weeks, because a foregoing study with this release system had shown a maintained tissue response for at least 24 weeks post-operation. After only 6 weeks, histological analyses revealed newly formed cartilage-like tissue, which resembled the adjacent, normal cartilage in cell as well as BMP-2 treated defects, but cell therapy gave higher histological scores. This advantage evened out until 12 weeks. Combinations of cells and BMP-2 did not result in any additive or synergistic effect. Equally efficient osteochondral defect repair was achieved with chondrocyte, stem cell, and BMP-2 treatment. Expression of collagen X and collagen I, signs of ongoing ossification, were histologically undetectable, and the presence of aggrecan protein indicated cartilage-like tissue. In conclusion, further work should demonstrate whether spatiotemporally controlled, on-site BMP-2 release alone could become a feasible therapeutic approach to repair large osteochondral defects.