<p>Landslides are catastrophic events that threaten life and property, particularly in rugged, high-slope mountainous regions. The Western Ghats (WG), the westerly escarpment along India’s western coast, are highly susceptible to landslides, especially during the downpours of the Indian monsoon. This study focuses on modelling landslide susceptibility in a small highland segment of Kerala, located, southern part of the Western Ghats. A landslide inventory map was prepared using multiple sources, documenting 260 landslides. Various conditioning factors, including Lithology, Land Use and Land Cover (LULC), Soil texture, distance from Streams, Topographic Wetness Index (TWI), Slope angle, Profile Curvature, Planform Curvature, Slope aspect, and distance from Roads, were used with the Random Forest machine learning technique for modelling. The model’s performance was evaluated using the Receiver Operating Characteristic-Area Under the Curve (ROC-AUC) metric. Results show that 4.85% of the total study area is highly susceptible to landslides, with 76.56% of the recorded landslides occurring in this region. 13.14% of the area is classified as having high susceptibility, contributing to 17.22% of the recorded landslides. Moderate, low, and very low susceptibility zones make up 31.80%, 27.61%, and 22.60% of the total area, respectively, with fewer landslides recorded in these regions. Besides, the model shows a ROC-AUC value of 0.913 and 0.903 during training and testing phases, respectively, indicating high model accuracy with minimal over-fitting. The study highlights the relative importance of various conditioning factors, with distance from roads, TWI, and slope angle being the most influential in determining landslide susceptibility. These findings provide important insights for land use planning, disaster management, and mitigation strategies.</p>