Extreme near-surface wind speeds in cities can have major societal impacts but are not well represented in climate models, in part due to coarse spatial resolution. Despite this, large-scale dynamics in the free troposphere, which models resolve better, could provide reliable constraints on local extreme winds. In this talk, I will discuss the synoptic circulations associated with midlatitude extreme wind events and assess how resolution affects their representation in reanalysis and a climate model framework. Composites of reanalysis (ERA5) sea level pressure and upper tropospheric winds during observed extreme wind events reveal distinct circulation structures for each quadrant of the surface-wind rose. I examine how these circulation patterns are represented in a climate model using the NCAR Community Earth System Model. A set of two otherwise identical simulations, one with globally uniform standard resolution (CESM-SE-UNIF), and the other with spatial resolution refined over Southern Ontario (VR-CESM), allows us to isolate the effects of regional refinement on the representation of extreme winds under historical climate conditions. Results show that VR-CESM improves representation of surface ageostrophic circulations and the strength of vertical coupling between upper-level and near-surface winds. Additionally, I will discuss preliminary work regarding simulations with end-of-century RCP8.5 forcing, which show diverging climate change projections of extreme winds between the uniform and refined resolution models. This work motivates further investigation into the added value of regionally refined resolution for the study of wind extremes under climate change.