Abstract Major element geochemistry of detrital garnet is a widely-used approach for sedimentary provenance analysis, in particular for unravelling parent-rock lithology of siliciclastic sediments. However, garnets from felsic crystalline rocks (e.g., intermediate–acidic igneous rocks and metapelites) often have similar major element composition. This results in ambiguity of provenance interpretation in most cases. Here, we collect trace element data of garnets from different rock types to explore the feasibility of detrital garnet trace element geochemistry in sediment provenance analysis. The Qaidam basin in northern Tibet is taken as an example and provenance of Cenozoic sedimentary rocks therein is interpreted based on detrital garnet trace element composition. We find that garnets in intermediate–acidic igneous rocks have much higher heavy rare earth element (ΣHREE > 600 ppm) and yttrium (Y > 800 ppm) abundances than garnets in metapelites (ΣHREE 4 ppm) and zinc (Zn > 150 ppm) abundances than garnets in low-, medium-grade metapelites (up to amphibolite facies). Based on these findings, we suggest that Fe- and Mn-rich detrital garnets in the Cenozoic sedimentary rocks from the northern Qaidam basin were probably derived from low-, medium-grade metapelites and Mg-rich, Ca-poor detrital garnets therein were most likely derived from granulite-facies metapelites. This provenance interpretation supports that the Qilian Mountains to the north was the major source for the northern Qaidam basin and major deformation style along the Altyn Tagh Fault to west have been dominated by large-amplitude lateral offset and extrusion, rather than crustal thickening and uplift during the early Cenozoic. This study emphasizes high potential of detrital garnet trace element geochemistry in provenance analysis and reconstruction of tectono-sedimentary evolution of clastic sedimentary basins.