Mammoth Cave National Park (MCNP) contains a diversity of forest types due to a complex mosaic of landform, rock types and land-use history. The point-centered quarter method was used to collect data for a forested vegetation classification. Stratified random sampling was done in each of the various site types found within the boundaries of MCNP. A classification based on the information available in the matrix of species importance values from each site type was constructed using two-way indicator species analysis (TWINSPAN). Geographic Information System (GIS) analysis was used to devise an automated vegetation mapping model that can be used to predict vegetation from environmental variables. A tool to assess the accuracy of model predictions was devised. The predicted vegetation map was stored within the GIS, and allows access to a variety of data associated with inventored, classifed, and predicted plant community types.