The study of spatial distribution of the indoor radon has assumed in the last years a lot of interest. The geostatistical techniques turn out to be particularly promising. The present work presents the results of a study where around 4000 indoor radon data from Veneto, Friuli Venezia-Giulia and Alto Adige, collected during the sampling campaigns performed in dwellings and in schools, have been analyzed. After the definition of the common data set, the study of the spatial distribution of the phenomenon has been performed by examining the experimental variograms. Declustering techniques have been applied. Predictive maps were defined by using simulation techniques; they allow to determine the probabilities of exceeding defined concentration levels, the 'radon-prone' areas. Systematic results regarding the validation of these maps are reported. This methodological study indicates how it is possible to understand the geographical variability of the phenomenon, trying to find out correlations among indoor radon, geological characteristics (i.e. lithology, morphology, tectonics, soil gas) and building-specific features, which can significantly influence radon concentrations.