Selected Late Tertiary and Quaternary lavas from the Aleutian localities of Adak, Great Sitkin, Cold Bay, and Amak have been studied. Bulk rock analyses as well as electron probe analyses of the major phases are presented. The lavas, basaltic andesites, and andesites are characterized by high $$SiO_{2}, Al_{2}O_{3}, K_{2}O, Sr$$, and Ba and low MgO, Ni, Cr, Co, and Yb (relative to ocean ridge tholeiites). Anorthite-rich plagioclase, magnetite, and large clinopyroxenes typify the phenocryst assemblage, while, in general, orthopyroxene occurs in those lavas possessing more than about 51.5 wt % $$SiO_{2}$$ and olivine replaces it in the less silicic lavas. No alkali feldspar, ilmenite, or hydrous phases occur in these lavas. A xenocryst of mantle-like olivine was analyzed in one lava. The limited phenocryst zoning and small phenocryst-groundmass compositional contrast suggests crystallization within perhaps 10 km of the earth's surface. The suggested sequence of crystallization is plagioclase, olivine and/or orthopyroxene, magnetite, and clinopyroxene, which is consistent with an oxygen fugacity similar to that denned by the Ni-NiO buffer. The plagioclase phenocrysts are relatively unzoned or slightly reversed in zoning, suggesting that crystallization took place under relatively high temperature (~1,200°C), low pressure, and essentially anhydrous conditions. Since addition of water to a melt greatly reduces the thermal stability of plagioclase and promotes the appearance of olivine (replacing orthopyroxene and becoming the liquidus phase) in even the more silica-rich lavas, the magma from which these lavas were produced probably contained less than 2.0 wt % water. This implies that hydrous melting of mantle peridotite is precluded as a source for these lavas. A more likely source is the subducted oceanic crust. The distinctly low $$TiO_{2}$$ contents typical of andesitic lavas may reflect the stability of rutile in the subducted eclogite assemblage under partial melting conditions. The well-known island arc relationship between $$K_{2}O$$ and depth to the Benioff zone may arise from the buffering of this component by sanidine in the eclogite assemblage.