Agricultural water demand in the Yellow River Basin (YRB) is increasingly shaped by climate change and human activities, posing challenges to sustainable water management. This study analyzed the temporal evolution and driving factors of crop water requirement (CWR) for four major crops-spring wheat, winter wheat, spring maize, and summer maize-using a dynamic crop coefficient model based on MODIS NDVI (2000-2020), combined with the FAO Penman-Monteith method. Pearson correlation and random forest model were employed to identify dominant climatic and anthropogenic influences. Future CWR trends (2021-2100) were projected under nine combined scenarios, integrating three CMIP6 climate pathways (SSP1-2.6, SSP2-4.5, SSP5-8.5) with three cropping area strategies (baseline, +15 %, -15 %). Finally, an assessment framework was established to evaluate agricultural water stress under future scenarios. Results show that the NDVI-based model effectively captures intra-seasonal crop variation and improves CWR estimation accuracy. From 2000-2020, per-unit CWR showed a significant increase for spring wheat and spring maize, whereas a decreasing trend was observed for summer maize. Human activities, especially the irrigated area, were the main driver of CWR change, surpassing climatic factors. Future projections indicate a significant upward trend in per-unit CWR (p < 0.001), with wheat being more sensitive to emission scenarios. Spatially, high-CWR zones are expected to shift from the arid northwest to the central and lower plains. Scenario combined high emission with planting expansion exhibit an approximately 16 % increase in mean annual CWR and result in the highest projected water stress. These findings provide a scientific basis for adaptive water governance and climate-resilient agricultural planning in large river basins.