On White Island, New Zealand, the intensified period of strombolian–volcanian and phreatomagmatic explosive activity that commenced in March 1977 led to eruption of unusually primitive, high-Mg andesites. These are Fo₈₀₋₉₃ olivine-saturated rocks that have MgO contents up to 10% (Mg# = 65–71) and SiO₂ of 56–58%. They have incompatible trace element characteristics that are typical of arc rocks. ⁸⁷Sr/⁸⁶Sr, ¹⁴³Nd/¹⁴⁴Nd and ¹⁷⁶Hf/¹⁷⁷Hf ratios (0.7049–0.7053, 0.51282–0.51266 and 0.28301–0.28298, respectively) are consistent with subducted sediment addition and/or crustal input but there is no clear correlation of either isotope ratio with MgO. The rocks have modest (3–10%) ²³⁸U excesses at low (²³⁰Th/²³²Th) ratios (0.697 to 0.722). ²²⁶Ra–²³⁰Th disequilibria are also restricted but, unusually, include both ²²⁶Ra excesses and deficits with (²²⁶Ra/²³⁰Th) = 0.94–1.07. (²¹⁰Pb/²²⁶Ra)₀ ranges from 0.98 to 1.52 requiring gas accumulation that may increase over time and with decreasing MgO. Sr/Y and Tb/Yb ratios are both low and relatively invariant at 8 and 0.3, respectively, and along with the ²³⁸U excesses preclude an origin in which residual garnet was involved. The occurrence of some ²²⁶Ra deficits suggests the presence of residual amphibole during partial melting for some samples. Rapid magma ascent (to preserve the ²²⁶Ra disequilibria) limits the amount of possible melt–wall rock interaction that might reduce source-derived Tb/Yb ratios and in the mantle or raise ⁸⁷Sr/⁸⁶Sr in the crust. The White Island high-Mg andesites did not form by partial melting of eclogite in the subducting Pacific plate. Their primitive, olivine-saturated compositions suggest that their source was peridotitic and experimental data suggest that equilibration with peridotite at low temperatures at 0.5–1.5 GPa and in the presence of elevated alkalis can reconcile the high SiO₂ and MgO of the rocks. This may reflect stalling of the magmas beneath the continental lithosphere for a few 1000 years.