Two weeks of high-frequency radar measurements collected at the Alderney Race are compared with the results of a three-dimensional fully coupled wave–current model. Spatial current measurements are rare in this site, otherwise well investigated through modelling. Thus, the radar measurements offer a unique opportunity to examine the spatial reliability of numerical results, and can help to improve our understanding of the complex currents in the area. Comparison of observed and modelled surface current velocities showed a good agreement between the methods, represented by root mean squared errors ranging from 14 to 40 cm s−1 and from 18 to 60 cm s−1 during neap and spring tides, respectively. Maximum errors were found in shallow regions with consistently high current velocities, represented by mean neap and spring magnitudes of 1.25 m s−1 and 2.7 m s−1, respectively. Part of the differences between modelled and observed surface currents in these areas are thought to derive from limitations in the k-epsilon turbulence model used to simulate vertical mixing, when the horizontal turbulent transport is high. In addition, radar radial currents showed increased variance over the same regions, and might also be contributing to the discrepancies found. Correlation analyses yielded magnitudes above 0.95 over the entire study area, with better agreement during spring than during neap tides, probably because of an increase in the phase lag between radar and model velocities during the latter.