3D-dust distributions, daily predicted by the Tel Aviv University Weather Research Center, are used by the Ministry of Environmental Protection to distinguish between natural and anthropogenic aerosols. Two different dust prediction systems were used to perform a sensitivity experiment and to compare the accuracy of the models. The first one was the original DREAM model with four dust particle size classes (0.7, 6.1, 18.0 and 38.0 μm); only the first two classes have a radius less than 10 μm and can be used in comparisons with PM10 measurements. The second model used was the modified DREAM model (DREAM-8) with a more detailed set of dust particle size classes with a radius of between 0.1 and ~7 μm (0.15, 0.25, 0.45, 0.78, 1.3, 2.2, 3.8, and 7.1 μm). The sensitivity experiment was carried out by using a quantitative comparison between PM10 measurements of surface dust concentrations and those predicted by models, for the high dust activity season in Israel, from February to June 2006. Quantitative comparisons showed that the use of eight particle size classes including four less than 1 μm in dust modelling, instead of only two classes, improves dust forecasts.