Extreme surface wind speeds in urban regions can cause significant damage to the built environment. Understanding how extreme winds, which are used in infrastructure design, may change under climate change is a potentially important consideration for climate change adaptation. Climate model projections of future extreme winds are highly uncertain, in part because models do not adequately capture turbulent boundary-layer processes at small scales. Nevertheless, large-scale dynamics in the free troposphere, well resolved by models, could provide reliable constraints on local extreme winds. To validate model representation of extreme near-surface winds, we investigate how reanalysis products at varying resolution represent wind variability and the timing of observed extreme winds at weather stations in Canadian cities. To investigate the large-scale dynamics, we produce composites of ERA5 fields during observed extremes and repeat for simulated extreme winds in the National Center for Atmospheric Research’s Community Earth System Model. Agreement between the model and reanalysis suggests that this model is representing the large-scale teleconnections associated with extreme near-surface wind speeds, a first step in building confidence that models are capturing the dynamics driving these extremes.