Agricultural modellers often need detailed soil profile data with which to run their models. We combine an extensive soil profile database with the Harmonized World Soil Database, a 30 arcsecond raster database of soil information worldwide, and describe a statistical process to identify representative soil profiles for each of its 188 distinct soil types at different spatial resolutions. We then outline a method to cluster the soils in the Harmonized World Soil Database to produce soil maps at coarser resolution, and we describe derived global soil maps at spatial resolutions of 5 and 10 arcmin, which may be more practical for some large-scale modelling studies. The derived data files allow a user to select any point or area on land and then to access the set of soil profiles pertaining to the mapping unit selected, which are available in a format suitable for use in modelling applications. In situations where the user has little or no other information about the soils in the region of study, the methods described can be used to produce plausible soil profile information based on the most up-to-date global soil map currently available.