Abstract: Environmental DNA (eDNA) metabarcoding has emerged as a powerful approach for assessing aquatic biodiversity; however, it remains limited by uncertainties in OTU-clustering thresholds and taxonomic assignment. The Yangtze River, China’s longest river system, harbors over 4 300 aquatic species, with fish constituting a dominant group in freshwater ecosystems. Due to their sensitivity to water quality, fish serve as reliable ecological indicators for monitoring environmental change. To improve the reference value of eDNA in studying Yangtze River fish diversity, DNA sequences of 320 species were collected from Mitofish and NCBI databases. Five commonly used eDNA fragments (12S-AcDBM07, 12S-Mifish-U, 12S-Tele02, 16S-Ac16s, COI-PS1) were extracted from three genes (12S RNA, 16S RNA, COI) for genetic distance calculations under thresholds of 0.03, 0.02, and 0.01. This aimed to evaluate identification efficiency and explore applications in fish diversity studies. Results show that at thresholds of 0.03, 0.02, and 0.01, 137, 108, and 56 fish species could not be identified using the five fragments. Among them, the protein-coding gene COI-PS1 exhibited the highest identification rate, reaching 53.44%, 60.63%, and 72.50% respectively. Among the five fragments, 12S-Tele02 with 12S-Mifish-U had the greatest overlap in identified species, with 98.54%, 100%, and 91.11% consistency across thresholds. When selecting fragment combinations, “12S-Tele02” plus “COI-PS1” identified the most species under all thresholds (180, 206, and 257 species), differing only by 3, 6, and 7 species from the results of using all five fragments. In summary, this study highlights that the relationship between interspecies difference thresholds and species identification must be carefully considered in applying eDNA for Yangtze River fish diversity, to avoid false positives in OTU annotations. For species that are hard to distinguish but of high conservation value, targeted primer design can enhance monitoring efficiency. Moreover, combining eDNA technology with traditional netting can provide a more comprehensive assessment of local aquatic biodiversity.