© 2015 Published by Elsevier B.V. Eruptions of voluminous 18O-depleted rhyolite provide the best evidence that the extreme conditions required to produce and accumulate huge volumes of felsic magma can occur in the upper 10 km of the crust. Mesoproterozoic bimodal volcanic sequences from the Talbot Sub-basin in central Australia contain possibly the world's most voluminous accumulation of 18O-depleted rhyolite. This volcanic system differs from the better known, but geochemically similar, Miocene Snake River Plain - Yellowstone Plateau of North America. Both systems witnessed 'super' sized eruptions from shallow crustal chambers, and produced 18O-depleted rhyolite. The Talbot system, however, accumulated over a much longer period (>30 Ma), at a single depositional centre, and from a magma with mantle-like isotopic compositions that contrast strongly with the isotopically evolved basement and country-rock compositions. Nevertheless, although the Talbot rhyolites are exclusively 18O-depleted, the unavoidable inference of an 18O-undepleted precursor requires high-temperature rejuvenation of crust in an upper-crustal chamber, and in this respect the evolution of the Talbot rhyolites and 18O-depleted rhyolites of the Snake River Plain - Yellowstone Plateau is very similar. However, instead of older crustal material, the primary upper-crustal source recycled into Talbot rhyolites was comagmatic (or nearly so) felsic rock itself derived from a contemporaneous juvenile basement hot-zone. Whereas giant low δ18O volcanic systems show that voluminous melting of upper crust can occur, our studies indicate that felsic magmas generated at lower crustal depths can also contribute significantly to the thermal and material budget of these systems. The requirement that very high-temperatures be achieved and sustained in the upper crust means that voluminous low δ18O magmatism is rare, primarily restricted to bimodal tholeiitic, high-K rhyolite (A-type) magmatic associations in highly attenuated lithosphere. In the case of the Talbot system, at least, our data suggest that an unusually hot pre-history might also be required to thermally prime the crust.