The Mozambique Ridge (MOZR) is one of several bathymetric highs formed in the South African gateway shortly after the breakup of the supercontinent Gondwana. Two major models have been proposed for its formation - volcanic plateau and continental raft. In order to gain new insights into the genesis of the Mozambique Ridge, R/V SONNE cruise SO232 carried out bathymetric mapping, seismic reflection studies and comprehensive rock sampling of the igneous plateau basement. In this study, geochemical data are presented for 51 dredged samples, confirming the volcanic origin of at least the upper (exposed) part of the plateau. The samples have DUPAL-like geochemical compositions with high initial 87Sr/86Sr (0.7024–0.7050), low initial 143Nd/144Nd (0.5123–0.5128) and low initial 176Hf/177Hf (0.2827–0.2831), and elevated initial 207Pb/204Pb and 208Pb/204Pb at a given 206Pb/204Pb (Δ7/4 = 2–16; Δ8/4 = 13–167). The geochemistry, however, is not consistent with exclusive derivation from an Indian MORB-type mantle source and requires a large contribution from at least two components. Ratios of fluid-immobile incompatible elements suggest the addition of an OIB-type mantle to the ambient upper mantle. The MOZR shares similar isotopic compositions similar to mixtures of sub-continental lithospheric mantle end members but also to long-lived, mantle-plume-related volcanic structures such as the Walvis Ridge, Discovery Seamounts and Shona hotspot track in the South Atlantic Ocean, which have been proposed to ascend from the African Large Low Shear Velocity Province (LLSVP), a possible source for DUPAL-type mantle located at the core-mantle boundary. Interestingly, the MOZR also overlaps compositionally with the nearby Karoo-Vestfjella Continental Flood Basalt province after filtering for the effect of interaction with the continental lithosphere. This geochemical similarity suggests that both volcanic provinces may be derived from a common deep source. Since a continuous hotspot track connecting the Karoo with the MOZR has not been found, there is some question about derivation of both provinces from the same plume. In conclusion, two possible models arise: (1) formation by a second mantle upwelling (blob or mantle plume), possibly reflecting a pulsating plume, or (2) melting of subcontinental lithospheric material transferred by channelized flow to the mid-ocean ridge shortly after continental break-up. Based on geological, geophysical and geochemical observations from this study and recent published literature, the mantle-plume model is favored.