Spontaneous fractures (i.e. caused by sudden loading and muscle contraction, not by trauma) represent a significant percentage of proximal femur fractures. They are particularly relevant as may occur in elderly (osteoporotic) subjects, but also in relation to epiphyseal prostheses. Despite its clinical and legal relevance, this type of fracture has seldom been investigated. Studies concerning spontaneous fractures are based on a variety of loading scenarios. There is no evidence, nor consensus on the most relevant loading scenario. The aim of this work was to develop and validate an experimental method to replicate spontaneous fractures in vitro based on clinically relevant loading. Primary goals were: (i) repeatability and reproducibility, (ii) clinical relevance. A validated numerical model was used to identify the most critical loading scenario that can lead to head-neck fractures, and to determine if it is necessary to include muscle forces when the head-neck region is under investigation. The numerical model indicated that the most relevant loading scenario is when the resultant joint force is applied to the head at 8 degrees from the diaphysis. Furthermore, it was found that it is not essential to include the muscles when investigating head-neck fractures. The experimental setup was consequently designed. The procedure included high-speed filming of the fracture event. Clinically relevant fracture modes were obtained on 10 cadaveric femurs. Failure load should be reported as a fraction of donor's body-weight to reduce variability. The proposed method can be used to investigate the reason and mechanism of failure of natural and operated proximal femurs.