During the first fifteen years of the twentieth century, Oxford-based Scottish geographer Andrew Herbertson constructed a framework for comprehending and categorising climate and its interrelations: natural regions. Along with a large circle of students and collaborators, Herbertson promoted natural regions as the conceptual keystone for geographical teaching and research. This article shows how natural regions theory conceived of climate as an object that was differently defined in different academic disciplines. Geography’s climate, according to Herbertson and his supporters, was defined by its relations with other spatially distributed phenomena rather than being the quantifiable and isolable entity of modern climatology. Building on recent work in the history of cartography foregrounding map use and reception, the article also argues that natural regions were products of particular modes of map reading, comparison, and synthesis. Although maps were arguably the most influential medium for communicating natural regions, they also proved limited as bearers of the multiscalar version of climate that Herbertson and his successors sought to convey. Finally, the article explains how natural regions and associated conceptions of climate came to be sidelined in the mid-twentieth century as geographers foregrounded human agency in region formation and adopted climatology’s definitions and analytical tools. Revisiting the life and death of theories of natural regions illuminates the contested significance of climate in the discipline of geography, and contributes to ongoing efforts to pluralise the history of climate sciences.