Summary While in clinical settings, bone mineral density measured by computed tomography (CT) remains the key indicator for bone fracture risk, there is an ongoing quest for more engineering mechanics‐based approaches for safety analyses of the skeleton. This calls for determination of suitable material properties from respective CT data, where the traditional approach consists of regression analyses between attenuation‐related grey values and mechanical properties. We here present a physics‐oriented approach, considering that elasticity and strength of bone tissue originate from the material microstructure and the mechanical properties of its elementary components. Firstly, we reconstruct the linear relation between the clinically accessible grey values making up a CT, and the X‐ray attenuation coefficients quantifying the intensity losses from which the image is actually reconstructed. Therefore, we combine X‐ray attenuation averaging at different length scales and over different tissues, with recently identified ‘universal’ composition characteristics of the latter. This gives access to both the normally non‐disclosed X‐ray energy employed in the CT‐device and to in vivo patient‐specific and location‐specific bone composition variables, such as voxel‐specific mass density, as well as collagen and mineral contents. The latter feed an experimentally validated multiscale elastoplastic model based on the hierarchical organization of bone. Corresponding elasticity maps across the organ enter a finite element simulation of a typical load case, and the resulting stress states are increased in a proportional fashion, so as to check the safety against ultimate material failure. In the young patient investigated, even normal physiological loading is probable to already imply plastic events associated with the hydrated mineral crystals in the bone ultrastructure, while the safety factor against failure is still as high as five. Copyright © 2016 John Wiley & Sons, Ltd. Clinical computed tomography and finite elements‐based failure risk assessment of bony structures remains one of the central desires in bone biomechanics. We here present a physics‐oriented approach, considering that elasticity and strength of bone tissue originate from the material microstructure and the mechanical properties of its elementary components. Using a clinical computed tomography of a vertebra, we explicitely introduce the hierarchical organisation of bone in the framework of multiscale elastoplasticity in order to retrieve patient and location‐specific safety factors against failure.