Railway induced vibrations can be amplified or attenuated as they propagate through the foundations, floors and spans of nearby buildings. This change in vibration level, known as the building's coupling loss, is often accounted for in vibration predictions using generic, frequency independent adjustment factors. In this paper, a detailed investigation is conducted to identify methods for determining the coupling loss in new build scenarios and to investigate its dependence on various system parameters. Using the approach of separate source, propagation and receiver terms, the coupling loss is characterised as the difference between vibration velocity levels measured inside and outside the building. Three different configurations of the railway-soil-building are explored: the case where both track and building are present; the case of a new building; and the case of a new railway. In situ experiments of a building subject to excitation from nearby train passages and hammer impacts on the track are used to quantify the building response. The influence of the location of the sensors in the free field and in the building (both within a single floor and across multiple storeys) is investigated. The findings from the experimental study supports separate characterisation of the source and receiver terms, thus demonstrating that the railway can be represented by a series of on-track impacts when determining the building's coupling loss. Using this representation, a 3D building model is used to investigate the effect of dynamic soil characteristics, surface foundation type, and building geometry on the building's response to railway induced vibrations.