A model for spatial prediction of landslide hazard was applied to a watershed affected by landslide events that occurred during the winter of 1995-96, following heavy rains, and snowmelt. Digital elevation data with 22.86 m x 22.86 m resolution was used for deriving topographic attributes used for modeling. The model is based on the combination of logistic regression and principal component analysis (PCA) of road related landslides and assumes that the meteorologic factors which caused failures in the past will cause failures in the future. This model was intended to provide forest land managers with additional information on landslide hazards associated with roads. The advantage of the model is that it allows rapid assessment of spatial correlation of the topographic attributes on large areas. A Geographic Information Systems derived map using the principal component analysis model delineates areas with relatively high probabilities of failure in the basin.