More widely used geostatistical methods for modeling distributed phenomena requires the evaluation of the quality of the product (maps) obtained by their application. The method of evaluation of uncertainty i.e. quality of the map was described on the example of a map of mean annual air temperature in Serbia for the period from the years 1991 to 2009 that was obtained from a relatively small number of samples for the whole country area (110 meteorological stations). The uncertainty of the map, obtained by kriging interpolation was evaluated by applying a Monte Carlo simulation modeling method. [Projekat Ministarstva nauke Republike Srbije, br. III47014, br. III43007 i br. TR 36009]