Ophiolites in the Indus–Yarlung Zangbo (IYZ) suture (southern Tibet) have been interpreted as remnants of the Neo-Tethyan lithosphere. However, the discovery of diamonds and super-reducing, ultra-high pressure (SuR-UHP) mineral assemblages (e.g., coesite after stishovite, olivine after wadsleyite, native metals, alloys, and moissanite) in some of these massifs and associated chromitites requires a re-evaluation of their origin and evolution. A new petrological and geochemical study of the Purang ophiolite in the western IYZ suture sheds new lights on these issues. The depleted harzburgites of the Purang massif have low modal contents of clinopyroxene (< 2%), and high Cr# [100*Cr³⁺/(Cr³⁺+Al³⁺)] in spinel (> 40 ~ 70) and pyroxenes (> 16 in orthopyroxene, and > 20 in clinopyroxene), suggesting high degrees of melt extraction (> 20%). These features are not consistent with formation in a (ultra-) slow-spreading mid-ocean ridge. These peridotites have high modal contents of orthopyroxene; this, and the extremely high Cr# of spinels in these peridotites, suggests modification in a subduction zone. The clinopyroxene-rich harzburgites and lherzolites contain rare spinel–pyroxene symplectites after garnet. Their clinopyroxenes have low MREE-to-HREE ratios ((Sm/Yb)N < 0.1) at relatively high HREE concentrations, and are Na-rich but Nd-poor. The relatively enrichment of Na but depletion of Nd in clinopyroxene cannot be explained by refertilization with MORB melts but are consistent with an origin from Na-rich subcontinental lithospheric mantle (SCLM). All lines of evidence suggest that these peridotites underwent initial melting in the stability field of garnet–facies peridotites, followed by additional melting in the spinel–facies mantle. Whole-rock Os isotopic compositions of the Purang peridotites give ancient TRD model ages (up to 1.3 Ga), indicating that the formation of these ancient depletion residues predated the opening of Neo-Tethyan Ocean. These observations, together with recent studies on other IYZ peridotites, suggest that the Purang peridotites are genetically unrelated to the associated mafic crust. Instead, they represent ancient SCLM domains, initially formed beneath a continental margin, and then modified by subduction, before they were incorporated into the Neo-Tethyan ocean basin. This model is consistent with the deep-mantle-recycling model for the presence of SuR-UHP phases in the IYZ ophiolites. The infiltration of MORB melts through these ancient depleted peridotites during their final exhumation in a (ultra-) slow-spreading center may have refertilized them to produce the clinopyroxene-rich peridotites.