Determining the contributions of climate change and human activities to runoff variation is important for water resource management. An improved double mass curve method (IDMC), which can distinguish the individual impact of each climatic factor (precipitation and potential evapotranspiration), is proposed in this study to attribute runoff variation to climate change and human activities. The runoff variation characteristics of four subregions of the Jinsha River Basin (JRB) were detected using statistical methods, and the contributions of climate change and human activities to runoff variation were comprehensively assessed using IDMC, Budyko-based elasticity method, and slope change ratio of cumulative quantity analysis. Results show that annual runoff had an increasing trend in the upstream, midstream, and downstream of JRB and a decreasing trend in Yalong River Basin (YRB) of JRB during 1966-2016. Changed points in 2004, 1986, 1996, and 2005 were detected for the upstream, midstream, and downstream of JRB and YRB, respectively; the study period was, therefore, partitioned into the base period and impacted period for each subregion. Suggested attribution methods gave consistent conclusions that climate change dominated the runoff variation in the upstream of JRB with an average contribution rate of 86.5%, whereas human activities were the major force for the runoff variation in the midstream and downstream of JRB and YRB with average contribution rates of 73.0%, 113.6% and 75.8%, respectively. Furthermore, precipitation caused the increase of runoff in the upstream and midstream of JRB and the reduction of runoff in the downstream of JRB and YRB. Potential evapotranspiration caused the reduction of runoff in the subregions of JRB except for the midstream of JRB. The findings of this study are meaningful for water resources management in the JRB.