Knowledge about the spatial distribution of NO 2 concentrations and its main contributors is beneficial for setting up air quality measurement plans, assessing exposure, or in licensing procedures, where background concentration levels are required. To produce air quality maps with reasonable spatial resolution, a suitable dispersion model is needed. The province of Styria with 16.400 km² and a population of 1.2 million is the second largest in Austria. While the northern part is dominated by the Alps (peak elevation: 2.995 m), the Southeast is much more flat with gentle hills and some larger plains. In order to take topographical effects on the pollutant dispersion properly into account, wind field libraries have been computed using the prognostic non-hydrostatic mesoscale model GRAMM (Graz Mesoscale Model). The chosen horizontal grid resolution of 300 m did not allow to model wind fields for the whole of Styria in one single run; instead the province has been divided into more than 20 overlapping sub-regions. Local observations of wind speed,-direction and estimated stability classes have been used as meteorological input. Subsequently quasi steady state wind fields have been computed and stored for later use in dispersion modelling utilizing the Lagrangian particle model GRAL (Graz Lagrangian Model). In order to capture strong NO 2 concentration gradients in the vicinity of roads, horizontal grid spacing was set 10 m in the dispersion calculations. Building effects on dispersion have been taken into account by applying a simple mass-conservative diagnostic flow field model implemented in GRAL. Modelled annual mean NO 2 concentrations compare well with observations, although in some areas unexpected patterns appear. In some cases these could be reasonably explained by wrong emissions. Apart from correctly assigning some emission sources in space (e.g. tractors, domestic heating), wind field simulations should be further improved. High resolution maps for PM 10 and B(a)P are foreseen in future, too.