Highlights: • The geochemistry of a Cretaceous Tongan fore-arc basalt (FAB) suite is reported. • The Tonga FAB suite is very similar to the Poya Terrane basalts of New Caledonia. • Similar geochemistry to IBM FAB but not associated with subduction initiation • Possibly a remnant of the hypothesized back-arc East New Caledonia Basin Abstract: The Tonga fore-arc preserves a complex history of subduction initiation, back-arc basin formation and arc volcanism which has extended from the Cretaceous to the present. In this paper, we discuss the geochemistry of a Cretaceous basalt/dolerite/gabbro suite recovered in two dredges from the Tonga fore-arc at ~ 19°S. The geochemistry of the Tonga fore-arc suite is unlike that of the uniformly depleted MORB basalts of the subducting Pacific Plate and therefore is unlikely to be accreted from Pacific Cretaceous crust. The ~ 102 Ma age obtained for one Tongan fore-arc dolerite is contemporaneous with a major phase of Cretaceous subduction-related volcanism, recorded both in detrital zircon age populations and associated volcanics from New Caledonia and New Zealand. We believe that the Tonga fore-arc basalts are a remnant of a hypothesized, once extensive Cretaceous back-arc basin, called the East New Caledonia Basin, which we propose to have existed from ~ 102 to 50 Ma. The allochthonous Poya Terrane of New Caledonia is geochemically very similar to the Tonga fore-arc basalts and represents a younger (~ 84–55 Ma) remnant of the same basin. Subduction-related Cretaceous volcanics from the SW Pacific, representing both arc and back-arc settings, all appear to have similar Zr/Nb values, suggesting a common mantle component in their petrogenesis. The Tonga fore-arc basalts are also similar to fore-arc basalts recovered from the Izu-Bonin-Mariana fore-arc, but unlike these basalts they are not associated with subduction initiation.