This study aims to assess and improve the Swedish forecast and information capabilities for ground-level ozone concentrations in ambient air. The assessment is based on a set of archived results from the Swedish operational chemical transport model MATCH and Swedish in-situ measurements of ozone covering the period of May 2008 to November 2010. The evaluation comprises two major activities: The first activity is an analysis of the overall model performance using standard statistical metrics suitable for longer time series. The second evaluation activity comprises in-detail analyses of the specific ozone episodes occurring in Sweden during the study period. In addition, trajectory modelling is used to investigate the meteorological conditions and transport patterns associated with those episodes. The evaluation of the model results shows that the model scores well according to standard evaluation criteria and confirms results of other studies in that the model easily meets the data quality requirements of the EU air quality directive 2008/50/EC. However, from an operational forecasting and information perspective it would be desirable to further improve the prediction of, in particular, high-level ozone episodes. Two different activities in our study are dedicated to the task of improving the forecast and information capabilities: The first activity tests the usefulness of statistical postprocessing of model results using regression techniques. The tests show promising results although the model performance during high-level ozone episodes is not improved. A limitation of our study is the relatively small archive of model data available for calibration andevaluation. Adaptive post-processing methods have not been tested in our study. The second activity aimed to improve ozone forecasting is a high-resolution model run for the year 2010. The higher reso-lution run gives slightly better results than the coarser operational model, which can be attributed to a better resolution of the physiography and thus certain physical and chemical processes. In particular, high-resolution simulations provide a more realisticrepresentation of the spatial ozone variation which is desirable for environmental assessments with a longer time horizon. However, from the perspective of operational ozone forecasting the increase in resolution cannot correct systematic problems such as an under-prediction of ozone if the source of ozone is non-local and the long-range transboundary transport is not correctly described by the European-scale model used as boundaries. Other potential sources of error are incomplete or erroneous emissions, representativeness issues, oversimplifications in the model’s physical or chemical processes, lacking data assimilation and initialization and oversimplifiedboundary conditions. While several of these issues are already addressed in current initiatives such as the EU FP7-project MACC, it is clear that further work will be needed during the coming years. Further work should also be invested in a better exploitation of the international developments within MACC and in the establishment of operational high-resolution air quality forecasts for Sweden, using boundary values from European-scale forecasts provided by theMACC-ensemble of regional air quality models.