The 1989-1990 eruption of Redoubt Volcano produced medium-K calc-alkaline andesite and dacite of limited compositional range (58.2-63.4% SiO2) and entrained quenched andesitic inclusions (55% SiO2) which bear chemical similarities to the rest of the ejecta. The earliest (December 15) magmas are pumiceous, often compositionally banded, and the majority is relatively mafic (< 59% SiO2). The most silicic magmas of the eruption are the late December to early January domes (up to 63.4% SiO2). Subsequent magmas formed domes and rare pumices which converge on 60% SiO2. Chemical variations among ejecta comprise tight, linear, two-component arrays for all elements for which the analytical uncertainty is much less than the compositional range. The two-component arrays are interpreted as mixing arrays between unrelated magmas because several of the arrays are at steep angles to the normal liquid line of descent. Additionally, the felsic endmember cannot be easily related to the mafic endmember by normal high-temperature igneous processes (e.g., the silicic endmember has higher Zr yet lower Hf than the mafic endmember). Also relative enrichments of highly incompatible elements are dramatically different across the arrays. The mixing event must have preceded eruption by a significant, yet unspecified amount of time because groundmass glass compositions are homogeneous for all post-December samples (Swanson et al., 1994-this volume), in spite of the whole-rock chemical diversity. This implies time for additional crystallization after the mixing event. Swanson et al. (1994-this volume) discuss evidence for a potentially different mixing event recorded only in December 15 magmas. Cognate cumulate xenoliths composed of pl+cpx+opx+hb+mt+melt were recovered from January and April deposits. These blocks differ from local batholithic country rock in their low concentrations of incompatible elements (e.g., Rb < 5 ppm vs 20-90 ppm, Ba < 150 ppm vs 300-2000 ppm) and low SiO2 (< 50 wt.% vs > 60 wt.%). They have Mg, Cr, Ni, Sc, and V contents higher than the andesites, but lower than Redoubt basalts and basaltic andesites. Thus, they may be crystallization products of andesites, but do not represent the cumulate residue of basalt fractionation. The xenoliths were probably derived from a shallow or intermediate crustal chamber. ?? 1994.