New investigations of the geology of Crater Lake National Park necessitate a reinterpretation of the eruptive history of Mount Mazama and of the formation of Crater Lake caldera. Mount Mazama consisted of a glaciated complex of overlapping shields and stratovolcanoes, each of which was probably active for a comparatively short interval. All the Mazama magmas apparently evolved within thermally and compositionally zoned crustal magma reservoirs, which reached their maximum volume and degree of differentiation in the climactic magma chamber ∼ 7000 yr B.P.

The history displayed in the caldera walls begins with construction of the andesitic Phantom Cone ∼ 400,000 yr B.P. Subsequently, at least 6 major centers erupted combinations of mafic andesite, andesite, or dacite before initiation of the Wisconsin Glaciation ∼ 75,000 yr B.P. Eruption of andesitic and dacitic lavas from 5 or more discrete centers, as well as an episode of dacitic pyroclastic activity, occurred until ∼ 50,000 yr B.P.; by that time, intermediate lava had been erupted at several short-lived vents. Concurrently, and probably during much of the Pleistocene, basaltic to mafic andesitic monogenetic vents built cinder cones and erupted local lava flows low on the flanks of Mount Mazama. Basaltic magma from one of these vents, Forgotten Crater, intercepted the margin of the zoned intermediate to silicic magmatic system and caused eruption of commingled andesitic and dacitic lava along a radial trend sometime between ∼ 22,000 and ∼ 30,000 yr B.P. Dacitic deposits between 22,000 and 50,000 yr old appear to record emplacement of domes high on the south slope. A line of silicic domes that may be between 22,000 and 30,000 yr old, northeast of and radial to the caldera, and a single dome on the north wall were probably fed by the same developing magma chamber as the dacitic lavas of the Forgotten Crater complex. The dacitic Palisade flow on the northeast wall is ∼ 25,000 yr old. These relatively silicic lavas commonly contain traces of hornblende and record early stages in the development of the climatic magma chamber.

Some 15,000 to 40,000 yr were apparently needed for development of the climactic magma chamber, which had begun to leak rhyodacitic magma by 7015 ± 45 yr B.P. Four rhyodacitic lava flows and associated tephras were emplaced from an arcuate array of vents north of the summit of Mount Mazama, during a period of ∼ 200 yr before the climactic eruption. The climactic eruption began 6845 ± 50 yr B.P. with voluminous airfall deposition from a high column, perhaps because ejection of ∼ 4−12 km³ of magma to form the lava flows and tephras depressurized the top of the system to the point where vesiculation at depth could sustain a Plinian column. Ejecta of this phase issued from a single vent north of the main Mazama edifice but within the area in which the caldera later formed. The Wineglass Welded Tuff of Williams (1942) is the proximal featheredge of thicker ash-flow deposits downslope to the north, northeast, and east of Mount Mazama and was deposited during the single-vent phase, after collapse of the high column, by ash flows that followed topographic depressions. Approximately 30 km³ of rhyodacitic magma were expelled before collapse of the roof of the magma chamber and inception of caldera formation ended the single-vent phase. Ash flows of the ensuing ring-vent phase erupted from multiple vents as the caldera collapsed. These ash flows surmounted virtually all topographic barriers, caused significant erosion, and produced voluminous deposits zoned from rhyodacite to mafic andesite. The entire climactic eruption and caldera formation were over before the youngest rhyodacitic lava flow had cooled completely, because all the climactic deposits are cut by fumaroles that originated within the underlying lava, and part of the flow oozed down the caldera wall.

A total of ∼ 51−59 km³ of magma was ejected in the precursory and climactic eruptions, and ∼ 40−52 km³ of Mount Mazama was lost by caldera formation. The spectacular compositional zonation shown by the climactic ejecta — rhyodacite followed by subordinate andesite and mafic andesite — reflects partial emptying of a zoned system, halted when the crystal-rich magma became too viscous for explosive fragmentation. This zonation was probably brought about by convective separation of low-density, evolved magma from underlying mafic magma. Confinement of postclimactic eruptive activity to the caldera attests to continuing existence of the Mazama magmatic system.