Between 2008 and 2011 a survey of radon (222Rn) was performed in schools of several districts of Southern Serbia. Some results have been published previously (Zunic et al., 2010; Carpentieri et al., 2011; Zunic et al., 2013). This article concentrates on the geographical distribution of the measured Rn concentrations. Applying geostatistical methods we generate "school radon maps" of expected mean concentrations and of estimated probabilities that a concentration threshold is exceeded. The resulting maps show a clearly structured spatial pattern which appears related to the geological background. In particular in areas with vulcanite and granitoid rocks, elevated radon (Rn) concentrations can be expected. The "school radon map" can therefore be considered as proxy to a map of the geogenic radon potential, and allows identification of radon-prone areas, i.e. areas in which higher Rn radon concentrations can be expected for natural reasons. It must be stressed that the "radon hazard", or potential risk, estimated this way, has to be distinguished from the actual radon risk, which is a function of exposure. This in turn may require (depending on the target variable which is supposed to measure risk) considering demographic and sociological reality, i.e. population density, distribution of building styles and living habits.