Extreme heat events are becoming increasingly common. With extreme heat exposure often necessitating prompt emergency medical service (EMS) response, the situation is critical as a rapid response is necessary to prevent severe disability or death. However, limited EMS capacity requires reallocation of resources to meet the increased demand for EMS. Although prior literature suggests a slight positive association between heat and EMS demand, studies have not investigated the spatial patterns of EMS demand during extreme heat. This study compares predictive models of EMS demand during extreme heat events across space. First, an unsupervised clustering approach was utilized to characterize EMS demand spatially to identify clusters with similar EMS demand and heat vulnerability characteristics. Finally, linear regression and geographically weighted regression (GWR) models were evaluated in predicting the spatial demand of EMS.