Soil erodibility is a critical parameter in erosion modeling, yet conventional RUSLE K-factors often overestimate soil loss due to the lack of local calibration, particularly in Mediterranean regions. This study develops a locally calibrated K-factor for the Oued Aoudour basin (Moroccan Rif) based on the plasticity index (PI) and lithological controls, with qualitative field support from a custom-built rainfall simulator. A logarithmic regression between K_observed and ln (PI) was established using 11 samples representing the main lithological units. After integrating a lithological factor (F_lith_capped) and a final empirical correction factor (0.603587229), the model performance improved from R² = 0.465 to R² = 0.9918. Leave-one-out cross-validation confirmed internal stability (R² = 0.985). The developed K-factor ranged from 0.0332 (calcareous soils) to 0.2662 (flysch soils), with estimated soil loss averaging 39 t/ha/year – a 32% improvement over the conventional K-factor (52 t/ha/year). Field measurements using the rainfall simulator (10–12 t/ha/year) provided qualitative support for the model’s improved relative performance. The proposed PI-based K-factor offers a simple, low-cost, and transferable approach for improving RUSLE-based erosion assessments in Mediterranean basins and data-scarce regions worldwide.