Forest fire seriously damage the Himalayan region's socio-ecological stability, particularly in Uttarakhand's Tehri Garhwal district. This study aimed to analyze the forest fire susceptibility (FFS) utilizing three different modelling approaches: Random Forest (RF), Bagging, and Logistic Regression (LR). Using R studio and SPSS, the models were run considering 900 fire sites and 22 geo-environmental drivers. Susceptibility was divided into five classes using Jenks natural breaks, which outperformed the other interval approaches in decreasing intra-class variance. The results indicate that RF models provided the most reliable spatial distribution, identifying 17.84% of the district as very high susceptibility, while the LR model tended to over-predict very high susceptibility zone (19.86%). Model validation via receiver operating characteristic (ROC) curves demonstrated exceptional predictive accuracy; the RF model achieved the highest Area Under Curve (AUC) of 0.977 (97.74%), followed by Bagging (0.966) and LR (0.866). Notably the inquiry revealed that 46% of the total area situated in high to very high susceptibility zones, particularly within the Devprayag, Ghansali, Jakhani-Dhar and Narendra Nagar regions. In order to safeguard the endangered tribes and ecosystems of the Tehri Garhwal, forest agencies and disaster management authorities can use the produced FFS maps as a high-precision decision-support tool to prioritize resources, improve early warning system and execute focused conservation strategies.