The mantle-crust boundary beneath oceanic spreading centres is a major chemical and thermal interface on Earth. Observations in ophiolites reveal that it is underlined by a dunitic transition zone (DTZ) that can reach a few hundred meters in thickness and host abundant chromitite ore bodies. The dunites have been deciphered as essentially mantle-derived in most ophiolitic massifs; that is, reactional residues of interactions between peridotite and percolating melt(s). Although both dunite and chromitite in ophiolites have been the focus of many studies, the reasons for their systematic association remain unclear. In this study we have explored the inclusion content of the chromite grains disseminated in the dunites from the DTZ exposed in the Maqsad area of the Oman ophiolite where a former asthenospheric diapir is exposed. Similarly to chromite in chromitite ore bodies, disseminated chromite grains in dunites contain a great diversity of silicate inclusions. Based on the major and minor element composition of 1794 single silicate inclusions in chromites from 285 samples of dunite and associated rocks in the DTZ, we infer that the disseminated chromites formed by a similar 'metallogenic' process to the chromitites, and that, as a whole, dunites from the DTZ actually represent the low-grade end-member of a single, giant ore body. The nature of the silicate inclusions (amphibole and mica among others) enclosed in chromite grains in dunites from the Maqsad DTZ precludes their crystallization from an anhydrous primitive basaltic melt, and rather calls for a crystallization from a melt hybrid between common mafic melts and more exotic Si-, Na- and volatile-rich fluids. The hybrid parent medium of both dunites and chromitites results from the interaction between an asthenospheric diapir (the mid-ocean ridge basalt source), and a colder, altered lithospheric lid and hydrothermal fluids responsible for this alteration. The excess silica in the hybrid melt is provided by the incongruent dissolution of enstatite from mantle harzburgite and/or from moderate degree of partial melting of the altered gabbroic crust. The chemical composition of the silicate inclusions is more variable when enclosed in the disseminated chromites than in the chromitites, suggesting a greater variability of melt and/or fluid fractions involved in the genesis of dunites than of chromite ores. Finally, the DTZ can be viewed as a metamorphic contact aureole between episodically rising asthenospheric diapirs and formerly accreted axial lithospheric lids. Our conclusion about the chicken and egg dilemma linking dunites and chromitites beneath oceanic spreading centres (i.e. do the chromitites form in response to the formation of dunites or conversely?) is that the mantle dunitization itself is a potential process for the release of Cr and its re-concentration as chromite ores, and that in turn the competition between orthopyroxene (+/- plagioclase) and chromite fractionation during this fluid-melt-peridotite reaction process is responsible for the great mineralogical and chemical variability of the DTZ dunites.