Mantle plumes are hot, buoyant upwellings that rise from Earth’s core-mantle-boundary (CMB) at ~2891 km depth to its surface1. As they ascend towards the lithosphere – our planet’s rigid outermost shell – the decreasing pressure facilitates decompression melting, resulting in surface volcanism that shows no obvious association with plate boundaries1-3. This so-called hotspot volcanism, commonly identified in oceanic regions, occurs less frequently within continental interiors due to the presence of a thicker lithosphere, which limits plume ascent and associated decompression melting4. However, given challenges in resolving mantle plumes with geophysical imaging5,6, it remains uncertain whether limited continental volcanism implies few sub-continental mantle plumes. Here, we present inter-disciplinary evidence revealing the first clear example of an amagmatic “ghost” plume in eastern Oman – the Dani plume. Despite lacking present-day surface volcanism7,8, this plume is robustly imaged using P- and S-wave arrival-time residuals from distant earthquakes, recorded by a dense regional seismic network. The imaged low-velocity structure is overlain by positive present-day residual topography9,10, in a region enigmatically uplifted since the late Eocene11,12 (~40 Ma). Our analyses of kinematic reconstructions demonstrate that asthenospheric flow associated with the Dani plume modified Indian-plate motion in the late Eocene, allowing us to bound the likely arrival time of this plume beneath the lithosphere. Our findings imply that other “ghost” plumes likely exist on Earth, for example beneath amagmatic topographic swells on the African continent, whilst there is strong evidence for their existence in the recent geological past13,14,15,16. Besides offering an approach to identify hidden continental plumes, both at the present-day and via Earth’s geological record, our study suggests that CMB heat-flux estimates should be revised upwards, with implications for thermal and core evolution models.