It was hypothesised that subchondral bone thickness, hardness and remodelling are influenced by exercise intensity, and by location within a joint. Dorsal carpal osteochondral injury is a major cause of lameness in horses undergoing high intensity training. This project aimed to determine the subchondral bone thickness, formation, resorption and hardness at sites with high and low incidence of pathology in 2 year-old horses undergoing 19 weeks high intensity treadmill training or low intensity exercise, and to compare these factors between exercise groups. Dorsal and palmar test sites were identified on radial, intermediate and third carpal articular surfaces after euthanasia. Adjacent osteochondral samples from each test site underwent histomorphometric analysis (for subchondral bone thickness, osteoid perimeter, osteoid seam width, eroded cavity area and eroded cement line surface length) and microhardness testing. Bone from horses undergoing high intensity training was thicker with a greater osteoid perimeter, and at individual sites had a smaller osteoid seam width and eroded cavity. Exercise-related differences were most marked at dorsal locations. Maximal differences in bone formation indices were observed at dorsal radial and medial third carpal locations. Overall subchondral bone from dorsal sites was thicker with a greater osteoid perimeter. Subchondral bone from dorsal sites was approximately 35% harder than bone from palmar sites. These results show topographical variations in subchondral bone structure, formation, resorption and material properties and a site-specific response to exercise. The maximal response to exercise was at high load sites with a clinical predisposition to injury. These findings indicate that the combined effect of exercise and local load variations within a joint may lead to maximal adaptive responses or overload of these responses at sites predisposed to injury. © 2001 Orthopaedic Research Society. Published by Elsevier Science Ltd. All rights reserved.