A rectangular graft of autogenous tibial periosteum was sutured (with its cambium layer facing into the joint) onto the base of a five by ten-millimeter full-thickness defect in the patellar groove of each of 143 adolescent and adult rabbits. The rabbits were managed postoperatively by either immobilization, intermittent active motion, continuous passive motion for two weeks, or continuous passive motion for four weeks. When the animals were killed four weeks postoperatively, the contour of the patellar groove had been restored in all of the rabbits in the group that had had four weeks of continuous passive motion, and the newly formed tissue in all of the defects in this group had the gross, histological, and histochemical appearance of smooth, intact hyaline articular cartilage. Histologically, the nature of the tissue that had formed, as well as its surface regularity, structural integrity, and bonding to the adjacent cartilage, were significantly better in the group that had had four weeks of continuous passive motion than in any of the other groups. The results were significantly worse when the orientation of the periosteal graft was reversed (that is, when it had been sutured into the defect with the cambium layer of the graft facing the subchondral bone rather than into the joint) or when no periosteal graft was used. Biochemical analyses revealed that, in the group that had had four weeks of continuous passive motion, the total hexosamine content, the levels of chondroitin sulphate and keratan sulphate, and the ratio of galactosamine to glucosamine were all comparable with the values for normal articular cartilage. In contrast, in the groups that were treated by immobilization, intermittent active motion, or two weeks of continuous passive motion, as well as in the adult rabbits, the content of the first three of these substances was significantly less than normal. In the groups that were treated by immobilization, intermittent active motion, or two weeks of continuous passive motion, 32 to 47 per cent of the total collagen was type II, while in the group that had had four weeks of continuous passive motion, 93 per cent of the total collagen was type II. These results demonstrate that, under the influence of continuous passive motion, free autogenous periosteal grafts can repair a large full-thickness defect in a joint surface by producing tissue that resembles articular cartilage grossly, histologically, and biochemically, and that contains predominantly type-II collagen.