A drastic change in plate tectonics and mantle convection occurred around 50 million years ago within the Pacific hemisphere. This is exemplified by the prominent bend in the Hawaiian–Emperor seamount change which seems to indicate that the Pacific Plate abruptly changed its direction from mostly motion to the north to the west. The Izu-Bonin-Mariana (IBM) and the Tonga-Kermadec subduction zones formed at the same time and investigators have long empirically linked IBM formation and Pacific Plate motion with the initiation of this vast subduction system pulling the Pacific to the west. I shed new light on this topic by describing results from ocean drilling in the IBM system and the southwest Pacific within the context of a new generation of global geodynamic models which realistically treats the mechanics of subduction and plate tectonics. Both an abrupt Pacific Plate motion change and a change in mantle plume dynamics have been proposed to account for the Hawaiian–Emperor Bend, but debates surround the relative contribution of the two mechanisms. I’ll describe plate reconstructions and high-resolution global dynamic models that quantify the amount of Pacific Plate motion change. The models show that ridge demise as well as Izu–Bonin–Mariana subduction initiation are incapable of causing a sudden change in plate motion, challenging the conventional hypothesis on the mechanisms of Pacific Plate motion change. Instead, Palaeocene intraoceanic subduction in the northern Pacific (long known from the geology of Kamchatka) exerts a northward pull on the Pacific Plate, while its Eocene demise leads to a sudden 30–35° change in plate motion, accounting for about half of the Hawaiian–Emperor Bend. I suggest the Pacific Plate motion change and hotspot drift due to plume dynamics could have contributed nearly equally to the formation of the Hawaiian Emperor Bend. Such a scenario is consistent with available constraints from global plate circuits, palaeomagnetic data and geodynamic models.