In this study, a new attribution analysis method based on large-scale hydrological modeling is developed. This method assesses the impacts of climate change and human activities on streamflow while taking uncertainties into consideration. Within the proposed general method framework, time series analysis methods are used to quantify the streamflow change; Then a large-scale Soil and Water Assessment Tool (SWAT) model is set up and the Differential Evolution Adaptive Metropolis (DREAM) algorithm is employed to approximate the posterior distributions of model parameters with Bayesian inference; Attribution analysis is further conducted with the improved rationale and the uncertainties translate from modeling are evaluated. The proposed attribution analysis method is then applied to a case study of the Upstream Yangtze River (UYZR) in China. Results reveal that: (1) Streamflow decreases significantly in the UYZR. Streamflow has various changing properties at small scales. (2) Precipitation, maximum temperature, windspeed and land use/cover change (LUCC) tend to decrease streamflow, while the minimum temperature and relative humidity tend to increase streamflow in the UYZR. Among the driving factors, precipitation has the most dominant influence. (3) Attribution uncertainty of precipitation has the largest absolute value and relative attribution uncertainties of driving factors behave differently in various river basins. These findings can help enhance the understanding of influences of climate change and human activities on streamflow and are also meaningful to adaptive water resources management.