Trabecular bone (TB) has the capability to adapt its mechanical and morphological properties when subjected to external loads. This process of bone adaptation is often modelled in silico by calculating mechanical stimuli in a finite-element (FE) model, representing TB tissue, subjected to various boundary conditions (BCs). Most of the studies modelling TB adaptation applied BCs to FE models of TB directly to their surfaces or via rigid plates. This approach resulted in distribution of mechanical stimuli in the models different from those in in situ TB. In addition, the “side-effect” artifact, expressed as reduction in stress values in peripheral trabeculae due to loss of their connectivity, was observed in these models [1]. To avoid these negative consequences, an approach of applying BCs via layer of TB was implemented earlier for measurement of apparent stiffness of TB and was demonstrated to improve the predicted results, as compared to the scenarios when BCs were applied directly to the analysis domain [2]. This study investigates influence of BCs applied via layer of TB on in silico modelling of bone adaptation to mechanical loading. For this purpose, cuboids of TB were extracted from high-resolution peripheral quantitative computed tomography (HR-pQCT) scans of human distal tibia and used to develop baseline FE models. BCs were applied to the FE models in two ways: (i) via layer of TB and (ii) directly to their surfaces. Mathematical model of bone adaptation implemented in Fortran code was used to regulate adaptation process in these models. The obtained simulated results were element-by-element compared with the corresponding follow-up models developed from the scans of the same participants after six months of physiological loading. The cuboids embedded into a layer of TB demonstrated higher correspondence with the follow-up models, as compared to the ones with BCs applied directly to their surfaces.