Résumé : This paper concerns the improvements of NO2 forecast due to satellite data assimilation. The Ozone Monitoring Instrument (OMI) aboard NASA Aura satellite provides observations of NO2 columns for air quality study. These satellite observations are assimilated, with the optimal-interpolation method, in an air quality model from Polyphemus, in order to improve NO2 forecasts in Europe. Good consistency is seen in the comparisons of model simulations, satellite data and ground observations before assimilation. The model results with and without assimilation are then compared with ground observations for evaluating the assimilation effects. It is found that in winter the errors between model data and ground observations have been reduced after assimilation, indicating a better NO2 forecast can be obtained using satellite observations. Such improvements are not found in summer, which is probably due to the shorter life time and higher temporal variability of NO2 in the warmer season.