Abstract. In the Mediterranean area, aerosols may originate from anthropogenic or natural emissions (biogenic, mineral dust, fire and sea salt) before undergoing complex chemistry. In case of a huge pollution event, it is important to know whether European pollution limits are exceeded and, if so, whether the pollution is due to anthropogenic or natural sources. In this study, the relative contribution of emissions to surface PM 10 , surface PM 2.5 and total aerosol optical depth (AOD) is quantified. For Europe and the Mediterranean regions and during the summer of 2012, the WRF and CHIMERE models are used to perform a sensitivity analysis on a 50 km resolution domain (from −10 • W to 40 • E and from 30 • N to 55 • N): one simulation with all sources (reference) and all others with one source removed. The reference simulation is compared to data from the AirBase network and two ChArMEx stations, and from the AERONET network and the MODIS satellite instrument, to quantify the ability of the model to reproduce the observations. It is shown that the correlation ranges from 0.19 to 0.57 for surface particulate matter and from 0.35 to 0.75 for AOD. For the summer of 2012, the model shows that the region is mainly influenced by aerosols due to mineral dust and anthropogenic emissions (62 and 19 %, respectively, of total surface PM 10 and 17 and 52 % of total surface PM 2.5 ). The western part of the Mediterranean is strongly influenced by mineral dust emissions (86 % for surface PM 10 and 44 % for PM 2.5 ), while anthropogenic emissions dominate in the northern Mediterranean basin (up to 75 % for PM 2.5 ). Fire emissions are more sporadic but may represent 20 % of surface PM 2.5 , on average, during the period near local sources. Sea salt mainly contributes for coastal sites (up to 29 %) and biogenic emissions mainly in central Europe (up to 20 %).The same analysis was undertaken for the number of daily exceedances of the European Union limit of 50 µg m −3 for PM 10 (over the stations), and for the number of daily exceedances of the WHO recommendation for PM 2.5 (25 µg m −3 ), over the western part of Europe and the central north. This number is generally overestimated by the model, particularly in the northern part of the domain, but exceedances are captured at the right time. Optimized contributions are computed with the observations, by subtracting the background bias at each station and the specific peak biases from the considered sources. These optimized contributions show that if natural sources such as mineral dust and fire events are particularly difficult to estimate, they were responsible exclusively for 35.9 and 0.7 %, respectively, of the exceedances for PM 10 during the summer of 2012. The PM 25 recommendation of 25 µg m −3 is exceeded in 21.1 % of the cases because of anthropogenic sources exclusively and in 0.02 % because of fires. The other exceedances are induced by a mixed contribution between mainly mineral dust (49.5-67 % for PM 10 exceedance contributions, 4.4-13.8 % for PM 2.5 ), anthropogenic so...