Five one-year-old immature swine were subjected to twelve months of exercise training. Four matched swine with no training served as controls. After they were killed, four-millimeter-wide strips of bone taken from the anterior, medial, posterior, and lateral quadrants of the central femoral diaphysis were subjected to four-point bending tests to failure. It was found that although exercise did not change the mechanical properties of the cortical bone, it resulted in significant increases in the averaged femoral cross-sectional properties: 17 per cent in cortical thickness, 23 per cent in cortical cross-sectional area, and 21 per cent and 27 per cent in maximum and minimum area moments of inertia, respectively. These changes were due primarily to reduction in the diameter of the medullary canal. The analyses of bone composition showed that the bone density and biochemical contents of the control and exercised animals were similar, but the total volume and the dry, ash, and calcium weights of the exercised bone were significantly higher. These combined results suggest that prolonged exercise has a significant effect on the quantity of the bone, but not on its quality. Clinical Relevance: It has long been recognized that stress deprivation from immobilization in plaster casts results in profound bone atrophy, and it is generally accepted that a minimum level of activity is necessary for homeostasis of bone. These results show that exercise at a level comparable to that prescribed in running fitness programs for humans (65 to 80 per cent of maximum heart rate) can not only maintain homeostasis, but produce actual hypertrophy of bone. This work further suggests the importance of graduated, prolonged, supervised rehabilitation programs in overcoming osteoporotic states.