Plagioclase phenocrysts in pre-historic andesites provide insight to the dynamics of magma formation and eruption at Taranaki volcano, New Zealand. The phenocryst population has a diversity of relic cores and a total in situ 87Sr/86Sr range of 0.70440–0.70486. Within-sample 87Sr/86Sr variations of 0.00018 to 0.00043 indicate that many phenocrysts are antecrysts and/or xenocrysts, derived from multiple crystal mush bodies. The Sr-isotopic differences in the phenocrysts of consecutive eruptions indicate that different magmas were tapped or formed on a centennial timescale. Most phenocrysts have multiple resorption/calcic regrowth zone(s) with elevated FeO* but invariant MgO zonation profiles. They likely record mafic melt inputs, and subsequent storage at elevated temperature caused re-equilibration of the Mg gradient. However, distinct rim types record different final pre/syn-eruptive magmatic conditions. Those in magmas erupted at 1030–1157 CE, 1290–1399 CE and 1780–1800 CE are characterised by resorption and calcic regrowth with sharp MgO and FeO* gradients. They record the entry of mafic melt into the system a few days or less before eruption based on Mg diffusion chronometry. In contrast, most phenocrysts erupted at 1755 CE, 1655 CE, and a few pre-1 ka events, have texturally uniform rims, compositionally consistent with closed-system crystallisation. This suggests alternating external and internal eruption triggers. Alternatively, the rate of magma reactivation via intrusion may dictate whether there was sufficient time for a mineralogical response to be recorded in part or all of the system. With respect to anticipating future eruptions, the plagioclase phenocrysts suggest multi-stage magma priming but rapid onset of the final trigger.