The variability of atmospheric dust mineralogy influences the impact of desert dust on the Earth's radiative budget and biogeochemical cycles. Until now, atmospheric transport models have assumed that dust was a constant homogeneous mixture, hence neglecting this variability. The lack of mineralogical data in arid areas prevented a better description of the atmospheric dust composition, and we propose here a new formulation to estimate the mineral content of arid surfaces on a global scale. First, we collect a Database of Arid Soil Surface Mineralogy for eight major minerals: quartz, feldspar, calcite, gypsum, illite, kaolinite, smectite, and hematite, both for the clay and silt fraction. On the basis of this, we formulate a Mean Mineralogical Table that relates classical soil types to surface mineralogy. We use this table and the geographical distribution of soil types given in the Food and Agriculture Organization Soil Map of the World to obtain the mineralogy of arid surfaces globally. In order to validate these results, we present a compilation of measured mineralogical composition of dust samples with identified sources. The correlation between observed dust mineralogy and those inferred from soil types in corresponding areas is between 0.70 and 0.94. We then calculate the maps of the single scattering albedo and of the ratio of infrared extinction to visible extinction for the erodible fraction of arid areas. Mineralogical maps presented here will be used in future studies with an emission scheme in a global transport model.