The 2012 Havre eruption evacuated a crystal-poor rhyolite (∼3–7% crystals) producing a volumetrically dominant (∼1.4 km3) pumice raft, as well as seafloor giant pumice (5–8%) and lavas (12–14%) at the vent (∼0.1 km3), both of which have subtly higher phenocryst contents. For crystal-poor rhyolites like the Havre pumice, it can often remain ambiguous as to whether the few phenocrysts present, in this case, plagioclase, orthopyroxene, clinopyroxene, Fe-Ti oxides ± quartz, are: (a) autocrysts crystallizing from the surrounding melt, (b) antecrysts being sourced from mush and the magma plumbing system, or (c) xenocrysts derived from source materials or chamber walls, or (d) possibly a combination of all of the above. In crystal-poor magmas, the few crystals present are strongly relied upon to constrain pre-eruptive conditions such as magmatic temperatures, pressures, water content and fO2. A detailed textural and compositional analysis combined with a range of equilibrium tests and rhyolite-MELTS modeling provide the basis for distinguishing autocrystic vs inherited crystal populations in the Havre eruption. An autocrystic mineral assemblage of andesine plagioclase, enstatite and Fe-Ti oxides constrains the pre-eruptive conditions of the Havre rhyolite magma: magmatic temperatures of 890 ± 27°C, crystallization pressures at 2–4 kbars, oxygen fugacity of NNO + 0.4 and water concentrations (5.6 ± 1.1 wt.%). Inherited phases not in equilibrium with the host melt composition are clinopyroxene, An-rich plagioclase (> An53) and quartz. Rhyolite-MELTs modeling indicates the clinopyroxene and quartz have most likely been sourced from cooler, silicic mush zones in the Havre magmatic system. This study demonstrates that even in crystal-poor rhyolites it cannot be assumed that all crystals are autocrystic and can be used to constrain pre-eruptive magmatic conditions.