Two new U-Pb dates (sensitive high-resolution ion microprobe on zircon) have an important bearing on the age of the Marinoan (Elatina) glaciation, a presumed global chronostratigraphic marker that has been previously poorly constrained in terms of its numerical age. In the Grassy Group of King Island, intermediate sills dated as 575 ± 3 Ma intrude an Elatina-equivalent diamictite (the Cottons Breccia), cap carbonate, and postglacial shale. The sills are locally vesicular, stratigraphically limited, probably intruded at shallow depth, and probably closely postdate the end of Marinoan glaciation. In the Togari Group of northwest Tasmania, a rhyodacite flow dated as 582 ± 4 Ma underlies the Croles Hill Diamictite, which is at least partly glaciogenic. No cap carbonate is known from the Croles Hill Diamictite, but in other respects its stratigraphic setting is similar to the Cottons Breccia. The two dates together support a significantly younger age (ca. 580 Ma) for the Marinoan glaciation than some previous estimates, and suggest correlation with the Gaskiers Formation of Newfoundland. The new data cannot exclude the possibility of a ca. 620 Ma age for the Marinoan glaciation, as suggested by recent evidence from outside Australia, but this would require a more complex and much less probable interpretation of the Tasmanian stratigraphic relationships.