Concentrations of 10Be (t 1 2 = 1.5 × 106 y) and 26A1 (t 1 2 = 0.72 × 106 y) have been determined by accelerator mass spectrometry (AMS) in a suite of quartz samples taken from sandstone boulders in several moraines in Arena Valley, a dry valley adjacent to the Taylor Glacier in the Quatermain Mountains, Southern Victoria Land, East Antarctica. These isotopes are produced in surficial quartz by cosmic ray spallation of O and Si. The concentrations in these samples ranged from 6.1 × 105 to 3.0 × 107 at g-1 for 10Be and from 9.4 × 106 to 1.2 × 108 at g-1 for 26A1, depending upon the extent of exposure at the surface. Analyses of 10Be in several samples from a single moraine at altitudes ranging from 1300 to 1650 m permitted an estimation of the air attenuation pathlength of cosmic ray neutrons; the value of 142+96 -41 g cm-2 is consistent with other published values. Production rates of 17+6 -4 at g-1 y-1 for 10Be and 113+54 -16 at g-1 y-1 for 26A1 at 1300 m and 87°S and a 26Al: 10Be production ratio of 6.5+1.3 -1.3 were calculated from the data. These values correspond to sea-level production rates at high geomagnetic latitude of 6.4 at g-1 y-1 and 41.7 at g-1 y-1 for 10Be and 26A1, respectively, consistent with determinations based on ~ 11 Ky glacially polished surfaces in the Sierra Nevada in California. These production rates imply exposure ages for the various moraines ranging from 50 Ky to 2.5 My, in accordance with other geological evidence. The 10Be and 26Al ages of these rocks compare favorably with those found using a similar dating method based on in situ production of 3He. Examination of the 3He concentrations in conjunction with the present data set yields an estimate of 3He production of 230+85 40 at g-1 y-1, corresponding to 85+31 -15 at g-1 y-1 at sea level. This is lower than previously reported values of 240 ± 60 at g-1 y-1 and ~190 at g-1 y-1 (scaled to sea-level and high geomagnetic latitude) based, respectively, on volcanic flows less than 2000 years old at low latitude, and samples collected from 14,400-year-old volcanic flows. This discrepancy is likely due to diffusive losses which may become significant (up to ~50%) for surface exposure ages on the order of 2.0 My. Nevertheless, the general correlations among the three isotopes suggest that 3He does not suffer gross losses through diffusion.