A 230Th-234U-238U dating study on pedogenic silica-carbonate clast rinds and matrix laminae from alluvium in Crater Flat, Nevada was conducted using small-sample thermal-ionization mass spectrometry (TIMS) analyses on a large suite of samples. Though the 232Th content of these soils is not particularly low (mostly 0.1-9 ppm), the high U content of the silica component (mostly 4-26 ppm) makes them particularly suitable for 230Th/U dating on single, 10 to 200 mg totally-digested samples using TIMS. We observed that (1) both micro- (within-rind) and macro-stratigraphic (mappabe deposit) order of the 230Th/U ages were preserved in all cases; (2) back-calculated initial 234U/238U fall in a restricted range (typically 1.67??0.19), so that 234U/238U ages with errors of about 100 kyr (2??) could be reliably determined for the oldest, 400 to 1000 ka rinds: and (3) though 13 of the samples were >350 ka, only three showed evidence for an open-system history, even though the sensitivity of such old samples to isotopic disruption is very high. An attempt to use leach-residue techniques to separate pedogenic from detrital U and Th failed, yielding corrupt 230Th/U ages. We conclude that 230Th/U ages determined from totally dissolved, multiple sub-mm size subsamples provide more reliable estimates of soil chronology than methods employing larger samples, chemical enhancement of 238U/232Th, or isochrons. Copyright ?? 2002 Elsevier Science Ltd.