Abstract: The runoff variation in the upper reaches of the Jinsha River is crucial to water resource management in the Yangtze River Basin. Its long-term changes are influenced by various climatic factors, thus it is necessary to analyze its driving factors and mechanisms. Based on station data, global sea surface temperature (SST) data, and reanalysis data, this study analyzed the variation and composition of runoff in the upper Jinsha River, and reveals the driving mechanisms of of Pacific and Atlantic SST anomalies on runoff during the key snowmelt period (April–June) and the main flood season (July–September). The results indicate that: (1) The annual average runoff in the upper Jinsha River shows a significant fluctuating upward trend; (2) The primary controlling factors of runoff recharge exhibit significant monthly differences. From April to June, the contribution of seasonal snowmelt to runoff increases significantly, while precipitation becomes the dominant factor controlling runoff variation at the Shigu Station from July to September; (3) The impact of SST anomalies on runoff displays regional and seasonal differences. In March, cold SST anomalies in the equatorial Pacific trigger an anomalous low-pressure system and warm advection over the upper Jinsha River, which significantly enhances precipitation and snowmelt during the snowmelt season, thereby increasing runoff; in June, the North Atlantic Tripole (NAT) SST anomalies regulate the Eurasian teleconnection wave train, intensifying southwesterly moisture transport and promoting increased precipitation during the main flood season, ultimately leading to higher runoff. These findings provide theoretical support for seasonal-scale runoff forecasting in the upper Jinsha River and offer a scientific basis for adaptive water resource management and flood/drought disaster prevention.