Both the Paris heat wave of 2003 and recent high-resolution climate change predictions indicate a world where mortality from extreme weather events will increase. Most heat wave deaths occur in buildings, and are driven by the thermal characteristics of the buildings and their local environment. Unfortunately previous work on the topic has ignored such spatial variations by either assuming the climate has little variation over a large area, or using archetypes of buildings from stock models. The latter forgetting that neither building characteristics nor landscape context are uniform over a city, with for example suburbs having a different architecture and shading to the inner city. In this work we use a statistical method combined with a new remote surveying tool to assemble accurate models of real buildings across a landscape then map the spatial variability in overheating and excess deaths now and in the future at a resolution of 5 km x 5 km. High spatial variation in the risk of overheating and heat-related mortality was found due to the variability of architecture, context and weather. Variability from the architecture and shading context were found to be a greater influence on the spatial variation in overheating than climate variability. Overheating risk was found to increase significantly with heat-related mortality tripling by the 2050s. The method was validated against data collected during the northern hemisphere 2006 hot summer. The maps produced would be a highly useful resource for government in identifying populations of greatest concern when developing policies to combat such deaths. (C) 2017 Elsevier Ltd. All rights reserved.