Rn-222 is commonly used as a natural tracer for validating climate models. Generally, a constant and homogenous Rn-222 source term of 1 atom cm-2 s-1 is assumed as a standard, sometimes reduced in northern latitudes. A tendency to overestimate measured Rn-222 concentrations by simulations with this standard assumption has often been found. To improve current models of atmospheric chemistry and transport a better source term for Rn-222 than currently used is necessary. This work aimed to establish a method for mapping the Rn-222 source term by using a commonly measured proxy, the terrestrial ?-dose (GDR) rate. A relatively stable fraction (~20%) of the total terrestrial GDR originates from the U-238 decay chain, of which Rn-222 is a member. In this study a regression model could be established by simultaneous measurements of Rn-222 flux and terrestrial GDR at locations in Switzerland and Germany. This model was validated on a regional scale by measurements in Finland and Hungary, at locations covering wide ranges of ?-dose rates. The predictions were within the error margin of measurements, and therefore considered to suffice to produce regional means of Rn-222 flux by using ?-dose rate as a proxy. To be able to develop a Rn-222 flux map for Europe, a base map for the ?-dose rate was necessary. For this instance, we used the large number of national ?-dose rate measurements, established after the nuclear reactor accident in Chernobyl in 1986. These data are composite values of terrestrial, cosmic and anthropogenic contributions and instrument background (self-effect). We extracted the terrestrial part of the total ?-dose rate provided by the EUropean Radiological Data Exchange Platform (EURDEP), which continuously udates and stores the data. Subsequently we produced annual, seasonal and weekly ?-dose rate maps for Europe (European Union, Norway, former Yugoslavia and Switzerland) with geostatistical methods. The regression model was then used to transform the terrestrial ?-dose rate maps into Rn-222 flux maps, using also additional information (organic/mineral soil, bare rock surface). Spatially and temporally resolved Rn-222 source maps for the European Continent resulted, with a spatial resolution of 0.5° x 0.5°. Previously made studies could be confirmed, and even more information was available now: modeled Rn-222 flux ranged from 0.03 to 1.76 atom cm-2 s-1, with a coefficient of variation of 51% and half of the values were between 0.40 and 0.70 atom cm-2 s-1. The weekly Rn-222 flux maps were applied in a simulation with the atmospheric transport model TM5, as well as the standard assumption of 1 atom cm-2 s-1 (with 0.5 atom cm-2 s-1 between 60°N and 70°N). The results from TM5 showed that our spatially resolved Rn-222 source term can improve predictions of atmospheric Rn-222 concentrations. In a case study in Gif-sur-Yvette (France) one week of Rn-222 concentrations were observed. The air mass trajectories turned (a) from areas with large (0.61 atom cm-2 s-1) to (b) areas with small (0.30 atom cm-2 s-1) Rn-222 fluxes. The standard assumption overpredicted atmospheric concentrations by (a) 70% and (b) 260%, while the simulation based on the new inventory followed the observation closely. On the basis of our approach we also produced Rn-222 flux maps for the United States of America and the Russian Federation territory, which are still preliminary and await verification.