This work investigates the hydrological aspects of desertification in drylands through water balance modelling. Rainfall, as the main water input to dryland hydrology, is characterised by strong variability over small spatial and temporal scales and determines water available as surface runoff, groundwater recharge and soil moisture. Thus, the focus here is on improving the realism of rainfall representation through optimal use of newly available remote sensing datasets. The hydrological importance of improved rainfall representation is investigated through modelling simulations using a dynamic, process-driven water balance model, DryMOD, applied in two distinctly different catchments in Tunisia and Senegal, respectively. The improved representation of rainfall in model-based scenarios helped to investigate small-scale variability of simulated rainfall, to identify runoff-generating areas, and to analyse the spatial distribution and seasonal variability of runoff and soil moisture ? all factors of paramount importance to water resource management in drylands.