Patient-specific simulations of bone remodelling could enable predicting how bone micro-structural integrity would be affected by bone diseases, drugs or other factors, and, ultimately could help clinicians to improve their prognoses. To simulate load-adaptive remodelling, however, knowledge about the physiological external loading acting on the bone is required. Assuming that load adaptation leads to homogeneous tissue loading, we previously developed a method to estimate the physiological loading history from bone micro-structural morphology. We were able to reconstruct the loading history of a simple load case that was applied in an animal experiment. However, we found considerable inhomogeneity in tissue loading suggesting that the bones were not fully adapted. Also, we noted differences in bone micro-architecture between animals despite common loading history, possibly due to differences caused by the stochastic nature of the bone remodelling process. In the present study, we aim at validating the load estimation algorithm in a well-controlled environment in which more complicated loading conditions are applied. Specifically, we want to test its accuracy for partially and fully developed bone structures and for differences in bone micro-architectures as they can occur due to stochastic events, even for bones with a common loading history. This was possible by using synthetic micro-architectures obtained from bone remodelling simulations as the basis for our load estimation algorithm. Loading histories based on fully adapted structures were predicted with a maximum error of 4.4% and predictions were not affected by differences in bone micro-architecture. These results show that our load estimation algorithm produces reasonable predictions and might be a suitable tool to define in vivo loading for patient-specific bone remodelling studies.