Atmospheric dust constitutes particles <100 μm, or deposits thereof (continental or marine); dust includes ‘loess,’ defined as continental aeolian silt (4–62.5 μm). Dust is well-known from Earth's near-time (mostly Quaternary) record, and recognized as a high-fidelity archive of climate, but remains under-recognized for deep time. Attributes such as thickness, grain size, magnetism, pedogenesis, and provenance of dust form valuable indicators of paleoclimate to constrain models of atmospheric dustiness. Additionally, dust acts as an agent of climate change via both direct and indirect effects on radiative forcing, and on productivity, and thus the biosphere and carbon cycling. Dust from the late Paleozoic of western equatorial Pangea reflects ultimate derivation from orogens (ancestral Rocky Mountains, Central Pangean Mountains), whereas dust from southwestern Pangea (Bolivia) reflects both proximal volcanism and crustal material. Records of dust conducive to cyclostratigraphic analysis, such as data on dust inputs from carbonate sections, or magnetism in paleo-loess, reveal dust cyclicity at Milankovitch timescales, but resolution is compromised if records are too brief, or irregular in interval or magnitude of the attribute being measured. Climate modeling enables identification of the primary regions of dust sourcing in deep time, and impacts of dust on radiative balance and biogeochemistry. Deep-time modeling remains preliminary, but is achievable, and indicates principal dust sources in the Pangean subtropics, with sources increasing during colder climates. Carbon cycle modeling suggests that glacial-phase dust increases stimulated extreme productivity, potentially increasing algal activity and perturbing ecosystem compositions of the late Paleozoic.