Summary The objective of this study is to assess recharge rates and their timing under layered loessial deposits at the edge of arid zones. Particularly, this study is focused on the case of the coastal plain of Israel and Gaza. First, results of a large-scale field infiltration test were used to calibrate the van Genuchten parameters of hydraulic properties of the loessial sediments using HYDRUS (2D/3D). Second, optimized soil hydraulic parameters were used by HYDRUS-1D to simulate the water balance of the sandy-loess sediments during a 25-year period (1990–2015) for three environmental conditions: bare soil, and soil with both sparse and dense natural vegetation. The best inverse parameter optimization run fitted the infiltration test data with the RMSE of 0.27 d (with respect to a moisture front arrival) and R 2 of 96%. The calibrated model indicates that hydraulic conductivities of the two soil horizons, namely sandy loam and sandy clay loam, are 81 cm/d and 17.5 cm/d, respectively. These values are significantly lower than those previously reported, based on numerical simulations, for the same site. HYDRUS-1D simulation of natural recharge under bare soil resulted in recharge estimates (to the aquifer) in the range of 21–93 mm/yr, with an average recharge of 63 mm/yr. Annual precipitation in the same period varied between 100 and 300 mm/yr, with an average of 185 mm/yr. For semi-stabilized dunes, with 26% of the soil surface covered by local shrub ( Artemisia monosperma ), the mean annual recharge was 28 mm. For the stabilized landscape, with as much as 50% vegetation coverage, it was only 2–3 mm/yr. In other words, loessial sediments can either be a source of significant recharge, or of no recharge at all, depending on the degree of vegetative cover. Additionally, the time lag between specific rainy seasons and corresponding recharge events at a depth of 22 m, increased from 2.5 to 5 years, and to about 20 years, respectively, with an increasing vegetative cover. For this reason, and also likely due to a great depth of loessial sediments, no correlation was found between annual recharge and annual precipitations of the same year or subsequent years. Similarly, no differences were found between summer and winter recharge fluxes. Instead, numerical simulations indicated continuous year-round recharge of the aquifer. We conclude that the layered subsurface acts as a short-term (annual) and long-term (multi-annual) buffer to smooth sudden precipitation/infiltration events. Vegetation conditions can help in predicting long-term recharge rates (as percentage of annual precipitation), which in turn need to be considered when assigning recharge characteristics in regional assessments and models.