Managers of wilderness resources must maintain, preserve, and sometimes restore pristine ecosystems while providing for public use and enjoyment of these areas. These managers require a resource information system that can store, retrieve and integrate basic data, synthesize components to solve particular problems, and provide simulations and predictions of natural processes and management actions. Traditional information systems based on land classification and type-mapping do not provide these capabilities. Gradient modeling, a new approach to resource management and forest fire simulation, has been developed to meet these needs in Glacier National Park. The method links four major components (1) a terrstrial site inventory coded from aerial photographs that offers 10-m resolution (2) gradient models of vegetation and fuel that derive quantitative stand compositional data from the parameters stored in the coded inventory, (3) a fuel moisture and microclimate model that extrapolates base-station weather data to remote sites using the parameters stored in the inventory; and (4) fire behavior and fire ecology models that integrate the data from the inventory and models to calculate real-time fire behavior and ecological succession following a fire.