Abstract. We examine the potential benefits of very high resolution for air-quality forecast simulations using a nested system of the Global Environmental Multiscale – Modelling Air-quality and Chemistry chemical transport model. We focus on simulations at 1 and 2.5 km grid-cell spacing for the same time period and domain (the industrial emissions region of the Athabasca oil sands). Standard grid cell to observation station pair analyses show no benefit to the higher-resolution simulation (and a degradation of performance for most metrics using this standard form of evaluation). However, when the evaluation methodology is modified, to include a search over equivalent representative regions surrounding the observation locations for the closest fit to the observations, the model simulation with the smaller grid-cell size had the better performance. While other sources of model error thus dominate net performance at these two resolutions, obscuring the potential benefits of higher-resolution modelling for forecasting purposes, the higher-resolution simulation shows promise in terms of better aiding localized chemical analysis of pollutant plumes, through better representation of plume maxima.