The last decade has seen frequent occurrences of severe air pollution episodes of high loading in SO2 during winters in the North China Plain (NCP). Using satellite data from the Ozone Monitoring Instrument (OMI), chemistry transport model (GEOS-Chem) simulations, and National Center for Environmental Predication (NCEP) meteorological reanalysis, this study examines meteorological and synoptic conditions associated with air pollution episodes during 2006e2015 winters. OMI-based SO2 data suggest a large decrease (~30% in area average) of SO2 emissions since 2010. Statistical analysis shows that meteorological conditions associated with the top 10% of OMI-based high SO2 days are found on average to be controlled by high pressure systems with 2 m s 1 lower wind speeds, slightly warmer, 1e2  C, temperatures and 10e20% higher relative humidities from the surface to 850 hPa. Numerical experiments with GOES-Chem nested grid simulations at 0.5  0.667 resolution are conducted for winters of 2009 as a control year, and 2012 and 2013 as years for sensitivity analysis. The experiments reveal that year-to-year change of winter columnar SO2 amounts and distributions in first order are linearly proportional to the change in SO2 emissions, regardless of the differences in meteorological conditions. In contrast, the surface SO2 amounts and distributions exhibit highly non-linear relationships with respect to the emissions and stronger dependence on the meteorological conditions. Longer data records of atmospheric SO2 from space combined with meteorological reanalysis are needed to further study the meteorological variations in air pollution events and the air pollution climatology in the context of climate change.