Multi-axis physiological loading is essential for evaluating bone-implant constructs, yet most experimental setups still apply simplified uniaxial or planar loading. This limits their ability to reproduce the complex 3D force and moment patterns occurring during daily activities. We present a new hexapod-based test system capable of applying controlled six-degree-of-freedom loading to bone-implant systems while simultaneously measuring forces, moments and interfragmentary mechanics. The testing device consists of a hexapod with flexible mold clamping system, equipped with two 6-axis force/moment sensors at the proximal and distal interfaces. A four-camera digital image correlation (DIC) setup records full-field surface displacements and strains across the fracture region. Artificial composite bones and human cadaveric specimens were prepared with standardized AO-classified fractures and stabilized using implants or fixators. Hexapod motions were programmed to reproduce kinematics representative of daily-life activities and partial weight-bearing. Time-resolved interfragmentary movement, local strain distributions, and global load paths were quantified and compared to ranges reported in mechanobiological literature for fracture healing. In parallel, specimen-specific finite element models with identical boundary conditions were used to validate the simulations against DIC and load cell measurements. The hexapod system reproduced complex multi-axis load paths, including combined axial loading, bending, torsion, and shear, with good repeatability of boundary forces and kinematics. DIC revealed heterogeneous strain fields and interfragmentary movement patterns that were dependent on fixation and fracture type. For selected loading conditions, measured interfragmentary strains fell within stimulus ranges reported in the literature, supporting the physiological relevance of the test protocols. FE predictions showed good agreement with experimental displacement and strain fields, confirming the suitability of the setup for validating numerical models. This provides a framework for implant design, loading-protocol optimization, and linking experimental data to mechanobiological healing models for future digital-twin applications.