<p>Groundwater recharge is critical because it restores the underground aquifers that supply billions of people. Over-abstraction of groundwater without recharge can lead to declining groundwater levels, land subsidence, and long-term ecological damage. As natural recharge is slow in many areas, particularly in arid and semi-arid climates, active management is required to ensure sustainability. Accurate estimation of groundwater recharge is crucial for sustainable water resource management, especially on islands where surface water resources are limited and vulnerable to climate variability. In this study, annual groundwater recharge in the island aquifer was estimated using two different approaches, the chloride mass balance (CMB) method and the numerical modeling tool HYDRUS-1D, using Bozcaada in the northern Aegean Sea, Türkiye. The CMB method is based on chloride concentrations measured in precipitation and groundwater samples, while the HYDRUS-1D model simulates water movement within the unsaturated zone over time. In this context, the study focuses specifically on recharge estimation challenges in small Mediterranean islands, where hydrogeological heterogeneity, limited monitoring networks, and marine influence complicate the reliable quantification of groundwater recharge. The analysis showed that the average annual recharge on the island was calculated to be 39 mm using the CMB method. In contrast, the HYDRUS-1D model provided more variable results for the different observation wells, with annual recharge estimated at 270 mm in the RK6 region, 75 mm in the RK12 region, and 165 mm in the RK13 region. The results indicate a higher recharge potential, especially in highly permeable formations such as the Kirazlı Formation, which includes sandstone. However, the study shows that the limitations of the CMB method become apparent on small islands and in areas at risk of seawater intrusion, leading to conflicting results between the methods. Consequently, it is emphasized that the combined use of approaches such as CMB and HYDRUS can provide a more comprehensive and reliable assessment, even under limited data conditions.</p> Graphical abstract <p></p>