Abstract: BackgroundTo identify the dominant driving factors of runoff and sediment variations in the mainstream of the Jinsha River basin, this study systematically investigates the characteristics of runoff and sediment as well as its influence factors based on long-term hydrological datasets during 1960-2022 from four hydrological stations: Zhimenda, Shigu, Panzhihua, and Pingshan. MethodsThis study employed Mann-Kendall non-parametric test, double cumulative curve method, cumulative slope method, and hydrological method to conduct trend detection and attribution analysis. Results①Runoff variation within the Jinsha River basin exhibited significant spatial heterogeneity. Specifically, significantly increasing trends were observed at the Zhimenda and Panzhihua stations, with change rates of 255 million m3/a (p<0.01) and 357 million m3/a (p<0.001), respectively. In contrast, the total sediment load showed a significant decreasing trend, with a rate of -408 million t/a (p<0.001). However, distinctvariations were found among individual stations: sediment at Shigu Station increased significantly (251 million t/a), while at Pingshan Station, it decreased significantly (-408 million t/a), and changes at other stations were not significant. ②The abrupt change of runoff generally occurred earlier than that in sediment load, with the timing of runoff transitions becoming progressively earlier from upstream to downstream. ③Attribution analysis indicated that climate change was the dominant factor driving runoff variations in the Jinsha River Basin (contribution rate>60%), whereas human activities were the primary cause of sediment reduction (contribution rate>84%). The dominant factors differ across different stations: increased runoff at Zhimenda was mainly attributed to climate change, but the rise in sediment load was primarily driven by human activities. At Shigu, both runoff and sediment changes are largely regulated by human activities (contribution rate > 65%). Conclusions The three attribution analysis methods show generally consistent results in quantifying the contribution rates of driving factors, but the cumulative slope method demonstrates relatively better robustness.