Bone provides stiffness, protects organs, and has metabolic functions such as fat and mineral storage or blood cell production. In their 3rd decade, humans reach their peak bone mass. An imbalance between bone formation and bone resorption leads to an increase in bone porosity with age. This increased porosity leads to an increased risk of fracture and, thus, induces pain, costs, and increased morbidity. The current gold standard, dual-energy X-ray absorptiometry (DXA), lacks accuracy in estimating bone strength and, thus, preventing fracture. Homogenized finite element (hFE) analysis based on high-resolution peripheral quantitative computed tomography (HR-pQCT) provides an accurate bone strength prediction compared to mechanical tests. However, the limits of the hFE methods still need to be determined. Therefore, the first part of this work is dedicated to investigating the achievements and limitations of hFE applied to distal tibia sections. At a lower scale, the hFE method is based on fabric-elasticity relationships. Although these relationships are known to hold for healthy conditions, it remains unclear if this bone mechanical adaptation capacity is preserved in pathological conditions. Therefore, fabric-elasticity relationships were investigated in patients diagnosed with osteogenesis imperfecta (OI). OI is a genetic bone disorder leading to dramatically increased fracture risks. The comparison between healthy and OI bone conditions has shown the hFE method’s suitability for OI bone conditions. At the tissue scale, cortical bone mechanical properties set in numerical simulations are often naive about age, sex, or anatomical location. Nevertheless, variations in mechanical properties of bone are linked to variations in its microstructure. Therefore, an automatic segmentation model was developed to quantify the microstructure of cortical bone at different, yet close, anatomical locations and on donors aged from 57 to 96 years old. Cortical bone microstructure does not differ with age or between sexes but presents significant differences between anatomical locations. This change in microstructure seems to be linked to mechanical strain experimented locally, overshadowing other factors such as sex and age. Altogether, these results provide new insights into bone biomechanics at different length scales from an experimental, numerical, and clinical aspect. Including hFE based on HR-pQCT scans in clinical routines should improve the overall bone health assessment compared to DXA. However, the hFE scheme is currently limited to predicting structural variables. It is suggested that a more morphological mesh be generated to improve the prediction of fracture zones and thus allow its use for vertebral fracture stability assessment. Additionally, the hFE scheme could be used in assessing OI patients’ bone health, and it is suggested to expand investigations to other pathologies affecting bone health, e.g., diabetes. However, due to its silent nature, osteoporosis remains challenging to diagnose before a fracture occurs. The most effective action to prevent fragility fractures might still be encouraging a healthier ageing with regular exercise, a balanced diet and balance training.