The Bolund experiment has been reproduced in a neutral boundary layer wind tunnel (WT) at scale 1:115 for two Reynolds numbers. All the results have been obtained for an incoming flow from the 270 o wind direction (transect B in the Bolund experiment jargon). Vertical scans of the velocity field are obtained using non-time resolved two components particle image velocimetry. Time-resolved velocity time series with a three component hot-wire probe have been also measured for transects at 2 and 5 m height and in the vertical transects at met masts M6, M3 and M8 locations. Special attention has been devoted to the detailed characterization of the inflow in order to reduce uncertainties in future comparisons with other physical and numerical simulations. Emphasis is placed on the analysis of spectral functions of the undisturbed flow and those of the flow above the island. The result's reproducibility and trustworthiness have been addressed through redundancy measurements using particle image velocimetry, two and three components hot-wire anemometry. The bias in the prediction of the mean speed is similar to the one reported during the Bolund experiment by the physical modellers. However, certain reduction of the bias in the estimation of the turbulent kinetic energy is achieved. The WT results of spectra and cosprectra have revealed a behaviour similar to the full-scale measurements in some relevant locations, showing that WT modelling can contribute to provide valid information about these important structural loading factors. Copyright © 2013 John Wiley & Sons, Ltd.