Soil moisture plays a critical role in land-atmosphere interactions and water resource variability over arid regions such as Egypt. However, the relative importance of soil moisture initialization versus land-surface physics in shaping simulated soil moisture profiles remains uncertain. In this study, we evaluate the sensitivity of soil moisture over Egypt within the RegCM4.7.0-CLM4.5 framework to different initialization strategies (bare soil, NOAA Century; CEN reanalysis, and ESACCI; ESA satellite product) and alternative runoff generation schemes (TOPMODEL and VIC). Soil moisture is analyzed at shallow (10 cm), intermediate (40 cm), and deep (100 cm) layers for the period 1998-2017 and evaluated against GLDAS-2.1.Results indicate that initialization exerts a dominant control on soil moisture magnitude and spatial distribution, with CEN initialization substantially reducing bias across all depths compared to bare and ESA initial states. Differences between runoff schemes primarily influence near-surface soil moisture, with VIC generally producing wetter conditions than TOP in selected regions. Application of linear scaling bias correction further improves shallow-layer representation and seasonal variability. Overall, the combination of CEN initialization and TOP runoff scheme yields the closest agreement with the reference dataset.These findings highlight the critical role of realistic soil moisture initialization in regional climate simulations over arid environments and provide guidance for improving soil moisture representation in climate-based water resource assessments.