In order to optimize the design of a national survey aimed to evaluate radon exposure of children in schools and to estimate radon distribution in Serbia, a pilot study was carried out in all the 334 primary schools of 13 municipalities of Southern Serbia. Rooms with annual radon concentration >300 Bq/m^3 were found in 5% of schools. The mean annual radon concentration weighted with the pupils is 73
Bq/m^3; weighted with the general population (assuming that the radon concentration distribution in dwellings is similar to that in schools) it is 94 Bq/m^3, both lower than the unweighted 119 Bq/m^3 average (–39% and –21%, respectively). Actual average concentration when children are in classrooms could be up to 95% lower. Variability between schools (CV=65%), between floors (CV=24%), and between rooms at the same floor (CV=21%) was analyzed. The impact of school location, floor, and room usage on radon concentration was also assessed (with similar results) by univariate and multivariate analyses. On average, radon concentration in towns is a factor of 0.60 lower than in villages, and at higher floors is a factor of 0.68 lower than ground floor. Results can be useful for other countries with similar soil and building characteristics.