The Cold Bay Volcanic Center has experienced two major stages of eruptive activity. Early (M-Series) acitivity produced bimodal Hi-Alumina basalt and calc-alkaline andesite lavas while later (FPK-Series) activity produced only calc-alkaline andesite. The spectrum of basalt compositions is believed to be due to high pressure (∼8 kb) fractionation at or near the base of the crust. Abundant mineralogical and geochemical evidence support a lower pressure mixing origin for all andesites. Inspection of the mineralogical data has shown that the earliest (M-Series) andesites were produced by mixing of basalt (<53 wt% SiO2) and silicic andesite (60.5 to 62.5 wt%) while later (FPK-Series) andesites resulted from the mixing of basaltic-andesite (53 to 56 wt%) and less silicic andesite (58.5 to 60.0 wt%). The major element and trace element geochemical data are consistent with a low pressure fractionation origin for the silicic endmember magmas and support the temporal variations in both mafic and silicic endmember compositions. The complete lack of crustal inclusions in all lavas is taken as evidence for a minimal crustal melting and/or assimilation role in the origin of the silicic endmembers. Many of the features of all andesites, including the important long term convergence of endmember magma compositions, are consistent with the process of liquid fractionation, accompanied by large scale magma mixing. A deduced upper limit of 62.5 wt% SiO2 for the silicic endmember magmas suggests that liquid fractionation, in the absence of major crustal melting, cannot produce more silicic magmas. A possible explanation is the presence of a rheological barrier, based on the concept of critical crystallinity (Marsh 1981), which prohibits more silicic liquids from being extracted from a crystal-liquid suspension.