Abstract: As a key tool for monitoring aquatic biological resources, environmental DNA (eDNA) technology faces challenges in quantitative biomass assessment due to multiple environmental factors, among which flow velocity is a critical one. Clarifying the effect of flow velocity on the degradation patterns of eDNA from cold-water fish can optimize the eDNA sampling methods and improve the precision of monitoring applications in stream environments. In this experiment, juvenile Schizothorax prenanti, an endemiccold-water economic fish species in the upper Yangtze River, were used as the target organism. An artificial annular flume was used to simulate a natural stream environment under three flow velocity gradients (0.2, 0.4 and 0.6 m/s), and the degradation dynamics of eDNA were investigated using quantitative Real-time PCR (RT-qPCR) combined with a monophasic exponential decay model. The results indicated that under different flow velocities, the eDNA of Schizothorax prenanti consistently exhibited a pattern of “rapid degradation in the early stage followed by persistent low-concentration retention in the later stage”, with positive signals still detectable after 28 days. Flow velocity showed a significant positive correlation with the eDNA degradation rate constant (K value); within the experimental range, higher flow velocity resulted in a higher degradation rate. The K values for the low, medium, and high flow velocity groups were 0.007/h, 0.021/h, and 0.033/h, respectively. The time required for 50% (T₅₀) and 95% (T₉₅) reduction in eDNA concentration in the low flow velocity group was 99.02h and 427.96h, respectively, which were significantly longer than those in the medium and high flow velocity groups. This study quantified the dynamic effects of flow velocity on eDNA degradation in a low-temperature stream environment and developed a flow-coupled eDNA decay model. The findings provide a scientific basis for optimizing field sampling protocols for rare and endemic fish species in the upper Yangtze River, and are expected to improve the accuracy of eDNA-based population abundance estimations.