The Beni Bousera peridotite massif, N. Morocco, contains pyroxenite layers of varied mineralogy, including graphitic garnet clinopyroxenite ( GGP ) layers. The graphite in these rocks occurs as octahedral multi-crystalline aggregates and other forms of cubic symmetry. Scanning electron microscopy and X-ray diffraction studies indicate that the graphite aggregates represent graphitized diamonds. Minor Na in coexisting garnets confirm ran origin in the diamond stability field. However, other mineral chemical data from the peridotites and pyroxenites indicate major sub-solidus re-equilibration of the silicate assemblages. Major and trace element analyses indicate the peridotites are residues of partial melting (5-30 %) in the spinel and possibly garnet stability fields. Some peridotites have suffered Fe and/or LREE enrichment. The pyroxenite suite crystallized as veins/dikes in the peridotites and show fractionation trends controlled by OPX, CPX, garnet and possibly spinel. The LREE depleted nature of the pyroxenite suite, combined with their highly variable incompatible element contents which do not correlate with fractionation indices such as Mg No., preclude their derivation from the host peridotites and require a chemically heterogeneous source/sources. Positive and negative Eu anomalies in the pyroxenites suggest their derivation from a low pressure precursor. Sr, Pb and Nd isotope analyses of the peridotites reveal large magnitude, small scale ( sub-km ) heterogeneities. Pyroxenites show even greater isotopic diversity and a decoupling of trace element-isotope systematics which indicate a recent melting event Sr, Pb and Nd isotopes indicate both peridotites and pyroxenites have experienced complex, long term evolution. The extreme isotopic diversity of the pyroxenites, with Pb isotope compositions that plot both to the left and right of the geochron, with high Δ7/4 and Δ8/4 values, are consistent with their derivation as melts of subducted oceanic crust plus less than 1% sediment, over 1 Ga ago. This interpretation is supported by pronounced oxygen isotope variability ( 8 ^ 18 0 = +4.9 to +9.3 %o ) suggesting the oceanic crustal source/s were hydrothermally altered before subduction. Hydrothermal alteration augmented Sr and Pb isotopicheterogeneity in the pyroxenites. If the isotopically light graphite in the GGP ( 8^13 C = -17 to -27 %o ) represents the carbon isotopic composition of the precursor diamonds, the original diamonds may have formed from subducted crustal ( kerogenous?) carbon. A model is proposed invoking subduction of hydrothermally altered oceanic crust and lithosphere, together with minor amounts of sediment (< 1 % ) into the asthenosphere over 1 Ga ago. The subducted oceanic slab descended to the 670km seismic discontinuity and "ponded” to form a megalith. Thermal equilibration of this megalith induced diapirism into the asthenosphere. Melting during ascent may have formed the pyroxenite layers and refertilized some of the peridotites. Asthenospheric upwelling during a major Neogene extensional event initiated emplacement of pyroxenite veined peridotitic mantle into the N. African, crust. Recent, small degree partial melting during ascent into the crust decoupled parent-daughter isotope systematics in the pyroxenites and peridotites. Graphitization of diamond in the GGP may also have occurred at this time. Pyroxenites containing graphitized diamonds in an orogenic peridotite massif provide evidence in support of a non-volcanic source for diamonds of unexplained provenance which occur in, or close to, major tectonic collision zones