Prediction of structure borne noise from industrial machinery allows not only to adopt mitigation measures to avoid excessive vibration at the design stage but also for problem solving after the final implementation. These measures also avoid potential damage to machinery tooling and the amount of defective products, whilst reducing the disturbance to operators within the working area and people in nearby facilities. Different strategies for achieving accurate predictions range from numerical and analytical methods, such as FEA and SEA, to methods which use measured data. Dynamic substructuring, an example of the latter, has been used for this research to characterise vibration contributions of three key assembly elements: source, isolator and receiver. The in-situ blocked force method was used to extract intrinsic properties of an operational vibration source as well as the mobility of the receiver and the dynamic stiffness of the isolators. Dynamic substructuring was then used together with the blocked forces data in order to predict structure-borne noise at the source interface with the isolators and at remote locations in the receiver. Predictions of source behaviour as part of an assembly were then compared with actual measurements made on the installation to serve as a validation of the method.