The Po Valley (Italy) model inter-comparison exercise (POMI) has been carried out in order to explore the changes in air quality in response to changes in emissions. The starting point was the evaluation of the simulated particulate matter and ozone (O-3) modelled concentrations against observations for the year 2005 of the six participating chemical transport models. As models were run with the same configuration in terms of spatial resolution, boundary condition, emissions and meteorology, the differences presented in the models' results are only related to their formulation. As described in the paper, significant efforts have been made to improve the accuracy of the anthropogenic emissions and meteorological input data. Nevertheless, none of the models using the proposed meteorology succeeded to fulfil the quality performance criteria set in the 2008 Air Quality Directive and in the literature for particulate matter, while also for ozone the results are not very satisfying. Although the overall performances look better for O-3 than for particulate matter with an aerodynamic diameter smaller than 10 mu m (PM10), the models tend to exhibit a similar behaviour and show the largest model variability in locations where concentrations are the highest (urban areas for PM10 and suburbs and hilly areas for O-3). While differences are significant in terms of standard deviation and bias, the correlation remains quite similar among models indicating that models generally capture well the main temporal variations, especially the seasonal ones. Possible explanations for this common behaviour and a discussion of the differences among models' results are presented in this paper.