Baseflow plays an important role in sustaining streamflow in the Yellow River watershed. Most of the recent researches have focused on the runoff variations in the Yellow River watershed, however, only a few studies investigated the variations of baseflow which are important to maintain the basic flow and ecological environment security. In this study, the Yanwachuan watershed was selected as the study area in the middle reaches of the Yellow River. Based on the hydrological, meteorological and NDVI (Normalized Difference Vegetation Index) data from 1981 to 2016 in the Yanwachuan watershed, nine numerical simulation methods were selected to separate the baseflow and were compared to analyze their applicability in the study area. The Mann-Kendall trend test and sliding t-test were then used to identify the changing trends and mutation points of the baseflow. Furthermore, the impacts of climate change and vegetation change on the baseflow were also qualitatively and quantitatively discussed. The results of this study can be drawn as follows:(1) Lyne-Hollick filtering method had higher accuracy than other separating methods, and its separating results were very consistent with the actual variations of daily baseflow. Thus, it was indicated that the Lyne-Hollick filtering method was more suitable for the baseflow estimation in the Yanwachuan watershed. (2) The average annual baseflow and baseflow index (the ratio of baseflow to runoff) of the Yanwachuan watershed were 0.152 m³/s and 0.58, respectively. According to the results of the M-K statistical test and sliding t-test, the annual baseflow and baseflow index were both significantly declined from 1981 to 2016 and their mutation points occurred in 1993 and 2006, respectively. (3) The correlation between baseflow and potential evapotranspiration was the strongest, while the correlation between baseflow index and the NDVI was the strongest and negative. The contribution rates of precipitation variation, potential evapotranspiration variation, and NDVI change to the decrease of baseflow were -99.1%, 113.3% and 85.8%, respectively, while the corresponding contribution rates to the reduction of baseflow index were 41.3%, -27.7% and 86.5%, respectively. It was clear that potential evapotranspiration variation and NDVI change were the main factors causing the decrease of baseflow, while the influence of NDVI change played a key role on baseflow index reduction.