Recent studies have shown that volumes of ancient depleted material can survive in the convecting asthenospheric mantle for long periods so that the use of Os model ages of mantle xenoliths to constrain the age of lithospheric mantle events should be approached with caution. In this study, we use in situ Os-isotope dating on sulfides in peridotitic xenoliths from cratonic (Tok, Russia) and off-craton (Tariat and Dariganga, Mongolia) settings of the Neoproterozoic-Phanerozoic Central Asia Orogenic Belt (CAOB) to examine lithosphere formation. A few Tok sulfides yield an apparent isochron indicating an age of 3.2 Ga. The high initial (super 187) Os/ (super 188) Os (0.117) of the apparent isochron suggests that it represents a mixing line, possibly involving an Archean component. In Tariat, both T (sub MA) ages from the least-disturbed sulfides ( (super 187) Re/ (super 188) Os<0.07) and T (sub RD) ages from higher-Re/Os sulfides yield model ages ranging from 0.5 to 3.0 Ga, with peaks around 1.7-1.5, 1.2 and 0.7-0.5 Ga. These ages suggest that SCLM beneath the Tariat region existed at least by Proterozoic time, and that some domains are Archean. The Os model ages are well-correlated with crustal events recorded in the overlying Precambrian Tarvagatay Terrane. It would be a remarkable coincidence if sulfides derived from randomly selected fragments of refractory materials in the convecting asthenospheric mantle would combine to give such a systematic correlation. We therefore prefer the simplest interpretation: the sulfide Os ages in the Tok and Tariat peridotites record major events that affected the crust+SCLM. The oldest of these events may record major melt extraction, and the later ones metasomatic events. Sulfides in Dariganga peridotites also have Mesoproterozoic Os model ages. Although Proterozoic crustal events have not been reported in this region so far, Proterozoic Nd model ages for basement rocks around the Xilinhot region in the vicinity of the Dariganga Plateau (B. Chen, pers. comm.) suggest that a Precambrian crustal terrain should be expected and might be found by studies of deep-crustal xenoliths in the Dariganga region.