Ash beds associated with the three climactic Yellowstone ignimbrites form important Quaternary chronostratigraphic markers over much of the continental United States. Previous K-Ar ages determined on crystal concentrates from these ashes varied by as much as 60–100 k.y. Laser-fusion 40Ar/39Ar dating of single sanidine grains from these units reveals a small number of grains with anomalously old ages. Eliminating these from the weighted averages results in highly precise refined ages of 2.003 ± 0.014, 1.293 ± 0.012, and 0.602 ± 0.004 Ma (2σ errors) for the Huckleberry Ridge Tuff, Mesa Falls Tuff, and member B of the Lava Creek Tuff, respectively. Individual single-grain ages that are slightly too old could result from incomplete degassing of xenocrysts in the magma. Electron-microprobe analyses of sanidine splits reveal no obvious xenocrystic compositions, suggesting another possibility—that phenocrysts from the crystallized rind of the magma chamber were re-entrained into the magma prior to eruption. Contamination and natural variation in phenocryst age may create larger uncertainty in bulk-crystal dating of young silicic volcanic rocks than incomplete extraction of Ar from sanidine.