Many bones within the axial and appendicular skeleton are subjected to repetitive, cyclic loading during the course of ordinary daily activities. If this repetitive loading is of sufficient magnitude or duration, fatigue failure of the bone tissue may result. In clinical orthopedics, trabecular fatigue fractures are observed as compressive stress fractures in the proximal femur, vertebrae, calcaneus and tibia, and are often preceded by buckling and bending of microstructural elements. However, the relative importance of bone density and architecture in the etiology of these fractures is poorly understood. The aim of the study was to investigate failure mechanisms of 3D trabecular bone using micro-computed tomography (microCT). Because of its nondestructive nature, microCT represents an ideal approach for performing not only static measurements of bone architecture but also dynamic measurements of failure initiation and propagation as well as damage accumulation. For the purpose of the study, a novel micro-compression device was devised to measure loaded trabecular bone specimens directly in a micro-tomographic system. The measurement window in the device was made of a radiolucent, highly stiff plastic to enable X-rays to penetrate the material. The micro-compressor has an outer diameter of 19 mm and a total length of 65 mm. The internal load chamber fits wet or dry bone specimens with maximal diameters of 9 mm and maximal lengths of 22 mm. For the actual measurement, first, the unloaded bone is measured in the microCT. Second, a load-displacement curve is recorded where the load is measured with an integrated mini-button load cell and the displacement is computed directly from the microCT scout-view. For each load case, a 3D snap-shot of the structure under load is taken providing 34 microm nominal resolution. Initial measurements included specimens from bovine tibiae and whale spine to investigate the influence of the structure type on the failure mechanism. In a rod-like type of architecture as seen in the whale spine, structural failure was described by an initial buckling and bending of structural elements followed by a collapse of the overloaded trabeculae. In the more plate-like bovine tibial architecture, buckling and bending could not be observed. Failure rather seemed to occur instantaneously. In conclusion, micro-compression in combination with 3D microCT allows visualization of failure initiation and propagation and monitoring of damage accumulation in a nondestructive way. We expect these findings to improve our understanding of the relative importance of density, architecture and load in the etiology of spontaneous fractures of the hip and the spine. Eventually, this improved understanding may lead to more successful approaches to the prevention of age-related fractures.