Using accelerator mass spectrometry (AMS), we have measured 10Be and 26Al in quartz from granites and sandstones from Table Mountain and Mount Fleming, Antarctica. Our data show that the plateau surface at Table Mountain had formed by early Pliocene time at the latest. Granites fringing but within the Sirius Group at Table Mountain give a minimum exposure age of 2.6 Ma for this deposit. A sandstone clast on the Ferrar dolerite surface just outside and below, and thus postdating, the Sirius Group has a minimum age of 2.9 Ma. Two samples from the Sirius Group at Mount Fleming have 10Be concentrations that have reached secular equilibrium. This deposit is at least 4.8 m.y. old. The Sirius Group at Mount Fleming cannot have been deposited after 3.0–2.5 Ma, as implied by biostratigraphic data. Our dates contradict the hypothesis that in the Pliocene East Antarctica was deglaciated and the climate was significantly warmer and wetter. The preservation of these surfaces indicates a continuous cold desert in the dry valleys since the beginning of the Pliocene. The high 10Be concentrations we have measured cannot be reconciled with uplift of the Transantarctic Mountains at a rate of 1 km/m.y. during the past 3 m.y.