Standard measurement methods of impact sound insulation of floor coverings (resilient surface layers) in laboratory are stated on Standard ISO 10140 series. Laboratory test allows evaluating the acoustical performance of resilient surface layers in terms of reduction of impact sound pressure level. Resilient surface layers applied on the top of the structural reference floor effectively reduce the impact noise produced in the receiving room by a tapping machine excitation. Anyway, an accurate characterization of the acoustical performances of the resilient surface layers depends on several boundary conditions. In this paper, it is shown a very relevant effect on the measurement accuracy, due to the cleaning condition of the structural reference floor. In particular it has been observed that the presence of sand and/or dust (in a very small quantity, i.e. of about 5 g/m^2) scattered on the bare slab greatly influences the experimental results. Researchers and technicians involved in building acoustic measurements in standard laboratories are well aware about the issues of proper cleaning, in particular if the resilient surface layers must be fixed on the surface by gluing. Nevertheless, the ISO 10140 standard (as well as previous ISO 140 standard series) does not state, as requirement, to glue the layers on the surface of the reference floor. As a consequence any accidental impurity on the bare floor surface can be a relevant source of inaccuracy.