Lameness in breeding-age gilts and sows is a major cause of culling, resulting in increased economic losses and welfare concerns. This study determined if exercise during gestation would affect the musculo-skeletal system, production variables, and behavior. Gilts were blocked by BW and assigned to 1 of 3 treatment groups: control (n = 10; no exercise), low exercise (n = 14; 122 m/d for 5 d/wk), and high exercise (n = 14; 122 m/d for 2 d/wk and 427 m/d for 3 d/wk). All gilts were stall-housed during gestation, and gilts were exercised between d 35 and 110 of gestation. Lameness score, BCS, BW, and blood were taken at multiple points before gestation, and during gestation and lactation. Blood serum was analyzed for carboxy-terminal telopeptide of type I collagen. Sow lying behavior was recorded for 3 d after farrowing. Farrowing data included litter weight and size at birth and weaning, and preweaning mortality. After weaning, 38 sows were slaughtered and muscles and the bones of the left fore- and hind-limbs were harvested. Bone density and quality were determined by computed tomography (CT) scans, dual energy x-ray scans, and bone-breaking force tests. The control group took longer to lie down than both exercise groups, and the low exercise group took longer to lie down than the high exercise group (P < 0.05). The number of pigs weaned was greater in the high exercise group than the control group (P < 0.05). Piglet preweaning mortality was greatest in the control group compared with both exercise groups (P < 0.05). The low exercise treatments exhibited a greater bone density (CT) in the humerus, radius, and tibia compared with that of the control group (P < 0.05). The bone density (CT) of the humerus in the low exercise group was greater than that of the high exercise group (P = 0.03). Breaking force in the humerus and femur was greater (P < 0.05) in the low exercise group than the control group. Breaking force in the tibia of the high exercise group was greater than the control group (P = 0.01). The tibia of both the low and high exercise groups had a greater breaking force (P < 0.05) than the control group. Although there was no benefit of exercise on lameness, differences in bone density and quality, lying behavior, and piglet survivability may provide useful insight into alternative housing for sows.