In view of the prominent problem of sharp runoff decline in the Yongding River Basin, we combined multiple hydrologic analysis methods of linear tendency, Mann-Kendall trend test and double accumulation curve with the complementary relationship conceptual model based on Budyko hypothesis to analyze the evolution trend of the Sanggan River Sub-basin and the Yanghe River Sub-basin from a qualitative and quantitative perspective, and to quantify the contribution rates of different factors to the runoff changes. The results showed that the runoff decreased significantly in the Sanggan River Sub-basin and the Yanghe River Sub-basin during 1957-2017, with the deceleration rate of 6.236 and 6.520 mm/10a and the abrupt change in 1969, 1983 and 1998, respectively. Precipitation decreased slightly; but both the temperature and NDVI increased significantly, with the acceleration rates of 0.321, 0.296 ℃/10a and 0.019/10a, 0.016/10a in the two sub-basins. The sensitivity coefficient analysis showed that the runoff of the two sub-basins was positively sensitive to the precipitation, and the sensitivity coefficients were 0.08 and 0.09 during the interference period; however it was negatively sensitive to the potential evapotranspiration, and the sensitivity coefficients were -0.03 and -0.02 during the interference period. The results of the complementary method based on Budyko hypothesis showed that the runoff of the Sanggan River Sub-baisn and Yanghe River Sub-basin decreased 27.5 and 22.3 mm respectively compared with those of the base period. The contribution rate of precipitation, potential evapotranspiration and change of the underlying surface to the Sanggan River runoff was 16.7%, 1.1%, 82.2%, and to the Yanghe River runoff was 7.2%, -4.5%, 97.3%, respectively. The change of underlying surface caused by human activity interference was mainly responsible for the decrease of runoff in the upper Yongding River Basin, compared with the Sanggan River Sub-baisn, the runoff in the Yanghe River Sub-baisn was more significantly affected by human activities.