The influence of vibrating buildings on the free-field ground motion could affect the earthquake recordings collected inside or nearby the buildings. Some evidences are known for large structures, but also small buildings could adversely affect the quality of the recordings. An example is given for a station of the Italian Accelerometric Network whose recordings show a clear mark of the frequency of the host building. To tackle this problem in a more general way, we performed numerical simulations whose first aim was to validate existing empirical evidence from a test site. Gallipoli et al. (Bull Seismol Soc Am 96:2457-2464, 2006) monitored a release test on a 2-storey R.C. building in Bagnoli (Italy), showing that a single vibrating building may affect the "free-field" motion with an influence that reaches 20% of peak ground acceleration. We re-analysed the data of that experiment following the Safak (Soil Dyn Earthq Eng 17:509-517, 1998) approach to building-soil motion, described as propagation of up- and down-going S-waves. The numerical model is a chain of single degree of freedom oscillators, whose dynamic behaviour depends on mass, stiffness and damping. The agreement between the synthetic and real data encouraged us to use this model to reproduce generalised structures as systems with a single degree of freedom. We run multiple tests varying the distance, between building and station, and the building-soil coupling, obtaining a statistical distribution of the influence of a single vibrating building on free-field ground motion taking into account the distance.