Effects of physical activity on bone properties, when controlled for genetic effects, are not fully understood. We aimed to study the association between long-term leisure time physical activity (LTPA) and bone properties using twin pairs known to be discordant for leisure time physical activity for at least 30 yr. Volumetric BMD and geometric properties were measured at the tibia shaft and distal end using pQCT in 16 middle-aged (50-74 yr) same-sex twin pairs (seven monozygotic [MZ] and nine dizygotic [DZ] pairs) selected from a population-based cohort. Paired differences between active and inactive co-twins were studied. Active members of MZ twin pairs had larger cortical bone cross-sectional area (intrapair difference: 8%, p = 0.006), thicker cortex (12%, p = 0.003), and greater moment of inertia (I(max), 20%, p = 0.024) at the tibia shaft than their inactive co-twins. At the distal tibia, trabecular BMD (12%, p = 0.050) and compressive strength index (18%, p = 0.038) were also higher in physically active MZ pair members than their inactive co-twins. The trends were similar, but less consistently so, in DZ pairs as in MZ pairs. Our genetically controlled study design shows that LTPA during adulthood strengthens bones in a site-specific manner, that is, the long bone shaft has a thicker cortex, and thus higher bending strength, whereas the distal bone has higher trabecular density and compressive strength. These results suggest that LTPA has a potential causal role in decreasing the long-term risk of osteoporosis and thus preventing osteoporotic fractures.