Wind speed and boundary layer height are key variables for studies of air pollution climatology and are routinely calculated by general circulation models in reanalysis and climate prediction projects. However, these models cannot yet resolve urban areas and the associated meteorological phenomena that are important for urban pollutant dispersion. Two simple air pollution models are examined; the box model and a Gaussian dispersion model. We find that, using meteorology from European Centre for Medium Range Weather Forecasts (ECMWF) reanalysis and operational analysis, the models most accurately reproduced the London climatology for daily mean concentrations of NO,, CO and PM10 from 1998 to 2005 if the boundary layer height was increased between 50 and 150m and high wind speeds were reduced. These corrections are consistent with the effects of the roughness and thermal properties of an urban area, which are not explicitly treated in the analyses. Perhaps more surprisingly, we find the best modelled pollution climatology by also specifying a minimum wind speed of at least 1 m s(-1). The minimum wind speed parameterises enhanced pollution ventilation under light wind conditions and could be consistent with an important role for the urban heat island circulation in reducing the severity of pollution events in London. These modifications are simple and computationally efficient, allowing offline comparison of pollution climates over long periods of time and potentially on a global scale. (c) 2008 Elsevier Ltd. All rights reserved.