Landslide susceptibility mapping is the first step in regional hazard management as it helps to understand the spatial distribution of the probability of slope failure in an area.An attempt is made to map the landslide susceptibility in Tevankarai Ar subwatershed,Kodaikkanal,India using binary logistic regression analysis.Geographic Information System is used to prepare the database of the predictor variables and landslide inventory map,which is used to build the spatial model of landslide susceptibility.The model describes the relationship between the dependent variable(presence and absence of landslide) and the independent variables selected for study(predictor variables) by the best fitting function.A forward stepwise logistic regression model using maximum likelihood estimation is used in the regression analysis.An inventory of 84 landslides and cells within a buffer distance of 10m around the landslide is used as the dependent variable.Relief,slope,aspect,plan curvature,profile curvature,land use,soil,topographic wetness index,proximity to roads and proximity to lineaments are taken as independent variables.The constant and the coefficient of the predictor variable retained by the regression model are used to calculate the probability of slope failure and analyze the effect of each predictor variable on landslide occurrence in thestudy area.The model shows that the most significant parameter contributing to landslides is slope.The other significant parameters are profile curvature,soil,road,wetness index and relief.The predictive logistic regression model is validated using temporal validation data-set of known landslide locations and shows an accuracy of 85.29 %. Landslide susceptibility mapping is the first step in regional hazard management as it helps to understand the spatial distribution of the probability of slope failure in an area. An attempt is made to map the landslide susceptibility in Tevankarai Ar subwatershed, Kodaikkanal, India using binary logistic regression analysis.Geographic Information System is used to prepare the database of the predictor variables and landslide inventory map, which is used to build the spatial model of landslide susceptibility. The model describes the relationship between the dependent variable (presence and absence of landslide) and the independent variables selected for study (predictor variables) by the best fitting function. A forward stepwise logistic regression model using maximum likelihood estimation is used in the regression analysis. An inventory of 84 landslides and cells within a buffer distance of 10m around the landslide is used as the dependent variable. Relief, slope, aspect, plan curvature, profile cu rvature, land use, soil, topographic wetness index, proximity to roads and proximity to lineaments are taken as independent variables.The constant and the coefficient of the predictor variable retained by the regression model are used to calculate the probability of slope failure and analyze the effect of each predictor variable on landslide occurrence in the study area. The model shows that the most significant parameter contributing to landslides is slope. the other significant parameters are profile curvature, soil, road, wetness index and relief.The predictive logistic regression model is validated using temporal validation data-set of known landslide locations and shows an accuracy of 85.29%.