Temperature–mortality analyses are challenging in rural and remote communities with small populations, but this information is needed for climate change and emergency planning. The geographic health areas of British Columbia, Canada were aggregated into four ecoregions delineated by microclimatic conditions. Time series models were used to estimate the effect of maximum apparent temperature on daily non-traumatic mortality. The population of the coldest ecoregion was most sensitive to hot weather, while the population of the hottest ecoregion was least sensitive. The effects were consistently strongest in decedents aged less than 75 years. A province-wide total of 815 deaths was attributed to hot weather over the 25-year study period, with 735 deaths in the most populous ecoregion. The framework described could be adapted to other climatically variable regions with urban, rural, and remote populations.