A suite of 23 basaltic to dacitic lavas erupted over the last 350 kyr from the Mount Adams volcanic field has been analyzed for U–Th isotope compositions to evaluate the roles of mantle versus crustal components during magma genesis. All of the lavas have (²³⁰Th/²³⁸U) > 1 and span a large range in (²³⁰Th/²³²Th) ratios, and most basalts have higher (²³⁰Th/²³²Th) ratios than andesites and dacites. Several of the lavas contain antecrysts (crystals of pre-existing material), yet internal U–Th mineral isochrons from six of seven lavas are indistinguishable from their eruption ages. This indicates a relatively brief period of time between crystal growth and eruption for most of the phenocrysts (olivine, clinopyroxene, plagioclase, magnetite) prior to eruption. One isochron gave a crystallization age that is ~ 20–25 ka older than its corresponding eruptive age, and is interpreted to reflect mixing of older and juvenile crystals or a protracted period of magma storage in the crust. Much of the eruptive volume since 350 ka consists of lavas that have small to moderate ²³⁰Th excesses (2–16%), which are likely inherited from melting of a garnet-bearing intraplate (“OIB-like”) mantle source. Following melt generation and subsequent migration through the upper mantle, most Mt. Adams magmas interacted with young, mafic lower crust, as indicated by ¹⁸⁷Os/¹⁸⁸Os ratios that are substantially more radiogenic than the mantle or those expected via mixing of subducted material and the mantle wedge. Moreover, Os–Th isotope variations suggest that unusually large ²³⁰Th excesses (25–48%) and high ¹⁸⁷Os/¹⁸⁸Os ratios in some peripheral lavas reflect assimilation of small degree partial melts of pre-Quaternary basement that had residual garnet or Al-rich clinopyroxene. Despite the isotopic evidence for lower crustal assimilation, these processes are not generally recorded in the erupted phenocrysts, indicating that the crystal record of the deep-level ‘cryptic’ processes has been decoupled from shallow-level crystallization.