The present article compares the outputs of the 3-D regional chemistry-transport model (CTM) CHIMERE against observations of the size-resolved aerosol chemical composition over northern Italy in clear sky and wind calm conditions. Two 4-day intensive field campaigns were carried out in July 2007 and February 2008 at three sites (urban, rural and mountain backgrounds) in the framework of the AEROCLOUDS project. Predicted levels are in reasonable agreement with observations for the urban and rural sites. Bias ranges from - 30%, for the rural site in winter, to + 38%, for the urban site during summer. In addition, the model is able to capture both the daily evolution of the bulk aerosol mass as well as its spatial gradients. Aerosol size distribution and chemical composition remain difficult to predict. The largest discrepancies were found for secondary organic aerosol (SOA) during summer and nitrates during the cold season. Compared with observations, modelled size distribution is shifted towards fine mode in winter, and towards coarse mode in summer. More accurate predictions can be achieved for both seasons by tuning the gas to particle absorption process. By reducing the SOA absorption rate by 25% at the urban sampling site in summer, the correlation between observed and simulated SOA size distributions increases from - 0.30 to + 0.70, and the bias is reduced from 200% to 0%. In winter, increasing the intra-sectional flux of particles from smaller to larger ones by a factor of 5, the Pearson correlation coefficient calculated over the nitrate size distribution goes up to + 0.85, compared to + 0.50 from CTRL, also resulting in a better agreement with the size distribution of PM10. As expected, the nitrate bulk mass concentration does not vary with respect to the base-case, and therefore nitrate overestimation remains present in the model.