To study mass transfers inside buildings equipped with ventilation systems, a methodology has been developed to carry out reduced-scale experiments for studying isothermal flows in a steady or a transient state. This methodology has been numerically and experimentally validated on simple configurations and applied to two reference configurations representative of nuclear facilities. The wind influence on mass transfers inside these configurations, in nominal, damaged (ventilation stopped) or accidental (internal overpressure) situations, has been studied by carrying out wind tunnel experiments. The objectives of this article are to present the scaling-down methodology and the main experimental results concerning the influence of wind on steady-state airflows inside the reduced-scale models. It is notably shown that wind effects can lead to a partial or a total loss of the pollutant containment inside buildings. The reliability of the zonal code SYLVIA, which is used to support safety assessments in nuclear facilities, has then been analysed from these experimental results.