Abstract Ignimbrite flare-ups are rare periods of intense silicic volcanism during which the pyroclastic volume and eruptive frequency is more than an order of magnitude higher than background activity. Investigating the compositional differences between flare-up and steady-state magmas provides critical constraints on the petrogenetic causes for the event and can offer unique opportunities to investigate the role of large-scale tectonic or geodynamic processes in arc magmatism. In this study, we focus on the bimodal Deschutes Formation ignimbrite flare-up of Central Oregon, which erupted unusually high volumes of pyroclastic material 6.25–5.45 Ma from a new axis of volcanism in the Cascades arc. This episode is marked by increased eruption rates and eruption of more silicic compositions relative to the Quaternary Cascade arc, which rarely erupts rhyolites. Ignimbrites are crystal-poor (<10%) dacite to rhyolites (mostly 65–77 wt.% SiO2) with anhydrous mineral assemblages and higher FeO/MgO, Y, Eu/Eu*, MREE and Zr/Sr, indicating drier magmatic evolution compared to the Quaternary arc, and are more similar to those from the rear-arc High Lava Plains (HLP) province that lies to the east. Magnetite-ilmenite oxybarometry indicates that Deschutes Formation felsic magmas tend to be hotter and more reduced (NNO-1 to NNO) than the Quaternary arc (NNO to NNO + 1.5). Rhyolite-MELTS geobarometry suggests complex storage of diverse Deschutes Formation magmas within the shallow crust (50–250 MPa), and the common co-eruption of multiple plagioclase populations, pumice compositions, and compositionally banded pumice suggest variable degrees of mixing and mingling of distinct magmas. Deschutes magmas also have low δ18Oplagioclase values that indicate partial melting and assimilation of hydrothermally altered shallow crust. Trace element systematics and rhyolite-MELTS modeling suggests that felsic pumice cannot be produced by simple fractionation of co-erupted mafic pumice or basaltic lavas, and requires a crustal melting origin, and trace elements and Pb isotopes suggest that young mafic crust may have been the primary protolith. We suggest that partial melting produced low-Si rhyolite melt (~72 wt.%) that acted as both a parent for the most evolved rhyolites, and as a mixing endmember to create the dacite to rhyodacite magmas with heterogenous plagioclase populations. Unlike the predominantly calc-alkaline basalts erupted in the Quaternary Cascade arc, Deschutes Formation primary basalts are mostly low-K tholeiites, indicative of decompression melting. These are similar to the compositions erupted during a contemporaneous pulse of low-K tholeiite volcanism across the whole HLP that reached into the Cascades rear-arc. We suggest that intra-arc extension focused decompression melts from the back-arc into the arc and that tensional stresses allowed this high flux of hot-dry-reduced basalt throughout the crustal column, causing partial melting of mafic protoliths and the production of hot-dry-reduced rhyolite melts. Depletion of incompatible elements in successive rhyolites implies progressive depletion in fertility of the protolith. Extension also allowed for the establishment of a robust hydrothermal system, and assimilation of hydrothermally-altered rocks by magmas residing in a shallow, complex storage network lead to low δ18O melts. Our findings suggest the integral role that extensional tectonics played in producing an unusual ignimbrite flare-up of hot-dry-reduced rhyolite magmas that are atypical of the Cascades arc and may be an important contributor to flare-ups at arcs worldwide.