Understanding exposure to air pollution during extreme events such as fire emergencies is critical for assessing their potential health impacts. However, air pollution emergencies often affect places without a network of air quality monitoring and characterising exposure retrospectively is methodologically challenging due to the complex behaviour of smoke and other air pollutants. Here we test the potential of roof cavity (attic) dust to act as a robust household-level exposure proxy, using a major air pollution event associated with a coal mine fire in the Latrobe Valley, Australia, as an illustrative study. To assess the relationship between roof cavity dust composition and mine fire exposure, we analysed the elemental and polycyclic aromatic hydrocarbon composition of roof cavity dust (<150μm) from 39 homes along a gradient of exposure to the mine fire plume. These homes were grouped into 12 zones along this exposure gradient: eight zones across Morwell, where mine fire impacts were greatest, and four in other Latrobe Valley towns at increasing distance from the fire. We identified two elements-barium and magnesium-as 'chemical markers' that show a clear and theoretically grounded relationship with the brown coal mine fire plume exposure. This relationship is robust to the influence of plausible confounders and contrasts with other, non-mine fire related elements, which showed distinct and varied distributional patterns. We conclude that targeted components of roof cavity dust can be a useful empirical marker of household exposure to severe air pollution events and their use could support epidemiological studies by providing spatially-resolved exposure estimates post-event.