<p>In Earth system modeling, plant rooting depth serves as a pivotal parameter that profoundly influences water resource allocation, runoff generation, deep drainage, and vegetation transpiration. Nevertheless, its significance in global-scale hydrological research has long been widely undervalued. Direct, large-scale observation of rooting depth remains a formidable challenge, primarily due to the inherent complexity of subsurface environments and limitations of current measurement techniques. Indirect estimation methods, such as inversion and optimization approaches, enable large-scale estimation of effective hydrological rooting depth and have thus emerged as indispensable tools for characterizing the spatial patterns of rooting depth. This study systematically evaluated and intercompared eight existing global rooting depth datasets (seven indirectly estimated datasets and one observation-based reference dataset). It further quantified the impacts of discrepancies among these datasets on hydrological processes simulations and identified key methodological limitations in previous rooting depth estimation research. Key findings reveal pronounced uncertainties in rooting depth estimates, which are particularly prominent in water-stressed regions. In these regions, variations in rooting depths can lead to remarkable discrepancies in hydrological simulations, with deviations of up to 40% in evapotranspiration, 31% in precipitation infiltration, and 51% in water yield. Mechanistic analysis demonstrates that the primary source of these uncertainties lies in the inadequate integration of root-soil water coupling mechanisms in existing modeling frameworks, particularly for water-stressed ecosystems. This study highlights the critical role of rooting depth in regulating terrestrial hydrological cycles and provides targeted recommendations for advancing rooting depth research, with a specific focus on addressing data scarcity and improving mechanism representation in water-stressed regions.</p>