Convective heat transfer at exterior building surfaces has an impact on the design and performance of building components such as double-skin facades, solar collectors, solar chimneys, ventilated photovoltaic arrays, etc. and also affects the thermal climate and cooling load in urban areas. In this study, an overview is given of existing correlations of the exterior convective heat transfer coefficient (CHTC) with the wind speed, indicating significant differences between these correlations. As an alternative to using existing correlations, the applicability of CFD to obtain forced CHTC correlations is evaluated, by considering a cubic building in an atmospheric boundary layer. Steady Reynolds-Averaged Navier-Stokes simulations are performed and, instead of the commonly used wall functions, low-Reynolds number modelling (LRNM) is used to model the boundary-layer region for reasons of improved accuracy. The flow field is found to become quasi independent of the Reynolds number at Reynolds numbers of about 10 5 . This allows limiting the wind speed at which the CHTC is evaluated and thus the grid resolution in the near-wall region, which significantly reduces the computational expense. The distribution of the power-law CHTC-U 10 correlation over the windward and leeward surfaces is presented (U 10 = reference wind speed at 10 m height). It is shown that these correlations can be accurately determined by simulations with relatively low wind speed values, which avoids the use of excessively fine grids for LRNM, and by using only two or three discrete wind speed values, which limits the required number of CFD simulations.