Direct perfusion of 3D tissue engineered constructs is known to enhance osteogenesis, which can be partly attributed to enhanced nutrient and waste transport. In addition flow mediated shear stresses are known to upregulate osteogenic differentiation and mineralization. A quantification of the hydrodynamic environment is therefore crucial to interpret and compare results of in vitro bioreactor experiments. This study aims to deal with the pitfalls of numerical model preparation of highly complex 3D bone scaffold structures and aims to provide more accurate wall shear stress (WSS) estimates. microCT imaging techniques were used to reconstruct the geometry of both a titanium (Ti) and a hydroxyapatite scaffold, starting from 430 images with a resolution of 8 microm. To tackle the tradeoff between model size and mesh resolution we selected two concentric regions of interest (cubes with a volume of 1 and 3.375 mm(3), respectively) for both scaffolds. A flow guidance in front of the real inlet surface of the scaffold was designed to mimic realistic inlet conditions. With a flow rate of 0.04 mL/min perfused through a 5 mm diameter scaffold at an inlet velocity of 33.95 microm/s we obtained average WSSs of 1.10 and 1.46 mPa for the 1 mm(3) and the 3.375 mm(3) model of the hydroxyapatite scaffold compared to 1.40 and 1.95 mPa for the 1 mm(3) model and the 3.375 mm(3) model of the Ti scaffold, showing the important influence of the scaffold micro-architecture heterogeneity and the proximity of boundaries. To assess that influence we selected cubic portions, of which the WSS data were analyzed, with the same size and the same location within both 1 and 3.375 mm(3) cubic models. Varying the size of the inner portions simultaneously in both model selections gives a quantification of the sensitivity to boundary neighborhood. This methodology allows to get more insight in the complex concept of tissue engineering and will likely help to understand and eventually improve the fluid-mechanical aspects.