The surface water resources of a typical semi-arid watershed (Huangfuchuan) in the middle reaches of the Yellow River have drastically decreased over the past decade, which has affected the governance strategies of the entire Yellow River. The causes of the decrease in surface water are generally attributed to climate fluctuations and human activities. In this study, a distributed biosphere hydrological model the Water and Energy Budget-based Distributed Hydrological Model (WEB-DHM) and a Contribution Assessment method were jointly applied to diagnose and quantify climate and human impacts on the streamflow change. Long-term hydrometeorological trends were analysed first and one major change-point (in 1998) in the annual streamflow series was identified through the nonparametric Mann–Kendall test and the annual precipitation-streamflow double cumulative curve method. The WEB-DHM model was calibrated and validated over the baseline period of 1985–1998; the natural streamflow was reconstructed for the impacted period of 1999–2006. Then, the contributions of climate fluctuations and human activities to streamflow change were determined quantitatively by comparing the natural streamflow with the observed value. The mean annual streamflow significantly decreased from 34.05 mm year−1 to 13.72 mm year−1 in the baseline and impacted periods, respectively, showing a reduction of 60%. Climate fluctuations accounted for a decrease in mean annual streamflow of approximately 10.38 mm year−1 (51.03%), whereas human activities (including soil–water conservation measures, artificial water intakes and man-made water storage infrastructure) caused a decrease of approximately 9.96 mm year−1 (48.97%). These findings are potentially helpful to support the water resources planning and management in the middle reaches of the Yellow River.