Samples of 17 volcanic rocks of Tertiary and Quaternary age from Adak and Kanaga islands have been chemically analyzed and studied microscopically. Spectrograms have been made of 10 of them. The rocks from Adak represent one center of possibly older Tertiary age and two centers of younger Tertiary or Quaternary age. The rocks from Kanaga Island represent both a shield volcano of possibly Tertiary age, partly destroyed by the formation of a caldera, and a young cone of Quaternary age that has grown within the caldera. All the rocks are basalt or andesite. Modally, all are characterized by relatively large crystals of plagioclase more calcic than andesine, and by one or more of the following ferromagnesian minerals: olivine, hypersthene, augite, and hornblende. Apatite and iron ores are common, and late silica minerals and orthoclase occur interstitially in the groundmasses of some rocks. As analyses of no more than four samples are available for each center, the small differences between sets of analyses representing different centers are of doubtful significance. Consequently, the analyses representing all the centers have been plotted on each of the several diagrams used. The several types of variation diagrams show that the province is a calc-alkaline one. The alkali-lime index is in the neighborhood of 63. This very high value is comparable with that for Katmai and is only slightly less than the maximum for the Japanese volcanic rock series. The quantities of minor constituents present are not exceptional for the rock types analyzed; the rocks from Adak are apparently more strontium-rich than those from Kanaga. The chemical analyses of the more basic rock types, as compared with the average analysis of plateau basalt, suggest that the Aleutian parental magma could have been derived from a plateau basalt magma by the addition of plagioclase and the subtraction of pyroxene, iron ore, and some quartz. The distribution of the minor elements can be explained more easily by postulating that, at least on Kanaga Island, some sediments have been assimilated. The derivation of the analyzed rocks from the Aleutian parental magma is most easily explained by the hypothesis that the plagioclase remained in suspension while the ferromagnesian minerals were settling out.