This article reviews interpretations of the geological and petrological nature of the Moho, which is defined as a discontinuity in terms of Vp, with a view to preparing for the Mohole on the ocean floor in IODP. We strongly propose discarding non-seismic terms for the Moho, such as “petrologic Moho”. The nature of the Moho has been controversial for a long time; an isochemical phase transition boundary between gabbro (crust) and eclogite (mantle) was favored for the Moho by some researchers, while a chemical boundary between mafic rocks (crust) and peridotite rocks (upper mantle) is now favored by a majority of researchers. Boundaries between completely or partially serpentinized peridotite and fresh peridotite may be applicable as the Moho at some parts of the ocean floors of a slow-spreading ridge origin. Antigorite serpentinite can be expected to be observed at the lowermost crust if the Moho is the serpentinization front at the stability limit of serpentine. The Moho beneath the Japan arcs can be estimated using mafic-ultramafic xenoliths in Cenozoic volcanics. Peridotitic rocks scarcely mix with feldspathic rocks, indicating that the Moho at that location is the boundary between feldspathic rocks (mostly mafic granulites; crust) and spinel pyroxenites (mantle). Possible fossil Mohos are observed in well-preserved ophiolites, such as the Oman ophiolite. Two types of Moho are distinct in the Oman ophiolite; gabbro-in-dunite Moho, where a gabbro band network in dunite changes upward to the layered gabbro within a few to several tens of meters, and dunite-in-gabbro Moho, where late-intrusive dunites intruded into gabbros. The former is of a primary origin at a fast-spreading ridge, and the latter is of a secondary origin at a subduction-zone setting in the obduction of the oceanic lithosphere as an ophiolite. The gabbro/peridotite (dunite) boundary as the primary Moho forms in embryo as a wall of melt conduit at fast-spreading ridges as well as at the segment center of slow-spreading ridges. The oceanic primary Moho is modified to various degrees by magmatism, metamorphism and tectonism in subsequent arc and continental environments. The gabbro-in-dunite Moho formation in the Oman ophiolite is an embryo of this modification. We expect in-situ sampling across the primary oceanic Moho formed at a fast-spreading ridge through the Mohole of IODP. Ultra-deep drilling at gabbro/peridotite complexes exposed on the ocean floor is indispensable for our understanding of the suboceanic upper mantle. Studies on appropriate ophiolites and deep-seated xenoliths from oceanic areas should complement the Mohole and other ultra-deep drillings to grasp the whole picture of the oceanic upper mantle.