In the analysis of machinery noise the aspect of sound source characterisation is of importance. Unlike for airborne sources, no widely applicable methods are available yet for structure-borne sound sources. In previous work a ‘pseudo-forces' methodology was suggested. In this approach fictitious forces on the outer surface replace the internal excitation in a source. The application of this approach is illustrated by experiments using a small air-compressor. In this example the pseudo-forces gave a good reconstruction of the response field (within 3 dB). Confidence intervals are derived for the results, which are 2–3 dB wide. This spread is separated into spread caused by inefficiencies of the method, spread caused by random measurement errors and the spread which is inherently generated by the source itself. These causes together enable the explaination of most of the deviations. Therefore, the pseudo-forces method is expected to serve well for particular structure-borne noise related questions.