<p>Most of the intermittent basins in the Mediterranean region lack gaging stations, making it difficult to estimate streamflow for effective water resources planning and management. In this study, we developed a methodology based on physical basins similarities for hydrological model parameter regionalization from gauged (donor) to ungauged basins (receivers). The approach was applied to five Mediterranean basins: two gauged basins (Lebna and Chiba river basins), and three ungauged basins (Lhjar, Boulidin and Mlaabi river basins). The Soil and Water Assessment Tool (SWAT) model was calibrated and validated at the donor basin using observed streamflow records. Further, the model parameters of the hydrological response units (HRUs) from the donor basin are transferred to the receiver HRUs based on similarity between their physical attributes namely climate, topography, land use and soil. The model performance after parameter transfer was assessed for the gauged basin (NSE = 0.5, PBIAS = 10.5, KGE = 0.5 and R² = 0.5). For the ungauged basins, we compared the spatial and temporal distribution of soil moisture simulated by the SWAT model both in the donor and receiver basins, with the Normalized Difference Moisture Index (NDMI) product from remote sensing. A good agreement between the simulated soil moisture and the spatial distribution of the NDMI, at both spatial and temporal scales (Pearson’s coefficient &gt; 0.5 at a significance level &lt; 0.05), highlighted that model parameter regionalization based on physical similarity shows potential for streamflow prediction in ungauged basins. This method provides a reliable approach for water resources management in regions with limited data. It can improve the reliability of streamflow estimation and support water resources management in data-scarce regions.</p>