Although osteochondral defects (OCD) following trauma, sport or degenerative diseases occur frequently, healing remains an unresolved clinical problem. These defects seem to appear more often in convex surfaces than in concave ones. In vivo studies have demonstrated the influence of mechanical conditions on osteochondral repair[ To study spontaneous repair, OCD (mm, 1.5mm depth) in 18 minipigs were created. Based on this knowledge, a predictive biphasic finite element model for tissue differentiation was created to simulate osteochondral healing. The model was validated by comparison of simulated healing with histological and histomorphometrical outcomes. Differentiation was regulated by the combination of a mechanical stimulus with a factor for differentiation defined for each tissue. The mechanical conditions arising from different predesigned defect fillings have been evaluated: Grafts with 100% (P1) and 50% (P2) of the native subchondral bone stiffness were analyzed. The healing pattern was in general qualitatively comparable to the findings of a gross examination of the histological sections. Generally, the pattern appears to be almost independent of the joint curvature. More hyaline cartilage (HC) was formed in the concave model during simulated healing. The maximum percentage of HC during the simulations was smaller and occurred earlier in the one (27 vs. 40%). In vivo 33% of HC was registered in the 12th week[ Until today the more frequent occurrence of OCD at convex joint surfaces reported in the clinical practice has not been related to the local mechanical environment. This study is the first to demonstrate that this may be related to the mechanical stimulus for healing. In fact, during healing simulation HC formation was affected by changes in the joint surface curvature. A continuity of material properties in the layers under an OCD, which operates as basis for the newly formed cartilage, is important for the development of a tissue with adequate mechanical quality for load transmission. Indeed hyaline cartilage formation occurs earlier when P1 as when P2 was used. The use of a predictive tissue differentiation model allows a better understanding of the mechanical aspects of healing. Further analysis is however required before such algorithm may be applied in clinical cases. To consider mechanical factors affecting healing, appear to be of importance.