Benefiting from the rapid development of environmental DNA (eDNA) technologies, sedimentary DNA (sedDNA) emerges as a promising tool for monitoring plant compositions in remote regions. The Tibetan Plateau (TP), renowned for its harsh environment and numerous ponds and lakes, presents a potentially demanding region for the application of sedDNA on vegetation investigations. Here, we used the g and h universal primers for the P6 loop region of the chloroplast trnL (UAA) intron to amplify plant DNA in surface sediments from 59 ponds and small lakes on the southwestern TP. The applicability and limitations of using plant DNA metabarcoding for modern vegetation monitoring and palaeo-vegetation reconstructions have been assessed by comparing sedDNA, pollen, and vegetation survey data. Our results showed that plant DNA metabarcoding recorded 186 terrestrial taxa, of which 30.1% can be identified at the species level. The plant sedDNA approach can effectively disclose the dominant plant taxa (including Asteraceae, Cyperaceae and Poaceae) and significant vegetation assemblages in the vicinity of the investigated sites. The number of taxa and taxonomic resolution of plant sedDNA exceeded that of pollen analysis (75 taxa detected, 5.3% can be identified at species level). Unlike pollen that retains a broad spectrum of regional plant signals (including Pinus and Artemisia), plant sedDNA mirrors very local plants, underscoring its utility in local vegetation monitoring and reconstructions. To conclude, plant DNA metabarcoding of (small) lake sediments warrant increased attention in the future for local vegetation monitoring and reconstructions on the TP.