Four autogenous osteochondral fragments removed from the lateral trochlear ridge of the talus were arthroscopically placed as loose bodies in a randomly selected middle carpal joint in each of 10 horses. The contralateral middle carpal joint, subjected to a sham procedure, served as control. Postoperative treatment was consistent with that for clinical arthroscopic patients. Lameness evaluation, radiographic examination, carpal circumference measurement, and synovial fluid analysis were performed before and at scheduled intervals after surgery. After a 2-month confinement, horses were subjected to an increasing level of exercise. Horses were euthanatized at intervals through 6 months. Gross and microscopic evaluations were performed on remaining fragments, articular cartilage, and synovial membrane of each middle carpal joint. Increased joint circumference, effusion, lameness, and degenerative joint disease distinguished implanted from control joints over the 6-month period. Implanted joints were characterized by grooved, excoriated cartilage surfaces, and synovium that was thick, erythematous, and irregular. At 4 weeks, implants were found to have adhered to synovium at their subchondral bone surface. The bone within fragments was undergoing necrosis, while cartilage was preserved. At 8 weeks, fragments were radiographically inapparent, grossly evident as pale plaques on the synovial surface, and composed of dense fibrous connective tissue. Synovial membrane specimens from implanted joints had inflammatory change characterized by mononuclear cell infiltration 2 months after implantation. Physical damage was apparent within articular cartilage of implanted joints at 2 months, and was significant (P less than 0.05) at 6 months after surgery. Chondrocyte degenerative change was significant (P less than 0.05) at 6 months after surgery. Focal reduction in safranin-O uptake was observed in cartilage layers adjacent to physical defects. Osteochondral loose bodies of the size implanted in the middle carpal joint of horses in this study were resorbed by the synovium within 2 months. Synovitis and significant articular cartilage damage were associated with the implanted fragments. Regardless of origin, free osteochondral fragments within the middle carpal joint should be removed, and methods to prevent residual postoperative debris should be implemented to reduce potential for articular pathologic change.