Soil organic carbon (SOC) is a key component of terrestrial ecosystems, serving as an energy source for soil microorganisms and playing an essential role in climate regulation and ecosystem productivity. However, SOC stocks are highly influenced by land-use and land-cover changes. This study aims to estimate and map SOC using multispectral Sentinel-2 and RapidEye imagery combined with environmental, soil, and topographic variables across Khat-dominated landscapes in the Haramaya district, eastern Ethiopia. A total of 88 soil samples were collected and analyzed in the laboratory for organic carbon estimation. Two machine learning algorithms, Random Forest (RF) and Extreme Gradient Boosting (XGB), were applied to predict SOC, and their performances were evaluated using the coefficient of determination (R²), root mean square error (RMSE and mean absolute error (MAE). Laboratory-measured SOC values ranged from 0.83% to 3.9%. Both satellite datasets produced comparable predictions, with Sentinel-2 estimating slightly higher mean SOC values (~40mg/ha), while RapidEye(~38mg/ha) provided more spatially detailed and accurate maps due to its finer resolution. Among the algorithms tested, RF outperformed XGB, showing higher predictive accuracy and stability, particularly under heterogeneous landscape conditions. The results suggest that both Sentinel-2 and RapidEye data are suitable for SOC estimation and mapping, with higher-resolution imagery preferred for detailed spatial analysis. Future research should focus on optimising predictor selection and assessing the potential impacts of Khat cultivation on SOC variability and spatial distribution.