This volume describes how ECOSEARCH species-habitat models are developed using information about the natural history of species, habitat descriptions, and specific descriptors for soils, topography, water regimes, and vertical habitat structure. Also, examples are provided that demonstrate the impacts of: 1) converting a wooded swamp into a recreational lake with adjoining golf course and housing lots in western Massachusetts; 2) converting a mature forest to earlier successional stages for woodcock management on a National Wildlife Refuge; and 3) varying logging practices on a national forest. ECOSEARCH was developed as a first step to a testable, scientific basis for ecosystem management. ECOSEARCH uses a FORTRAN program to predict occurrence for >300 species of New England wildlife using natural history models and information contained in geographicinformation matrices. ECOSEARCH is based on the assumptions that many wildlife species are associated with certain habitat attributes that can be estimated from vegetative structure, water regimes, soil types, and topographic conditions. The challenge of predicting species occurrences is to discover the constraints that structure particular ecosystems. The more rigid the constraint, the more reliable the prediction. Data from local landscapes are organized as raster data with 50-m pixels. At present, the vegetative-structure matrix is provided by interpreting a systematic grid of points from aerial photographs. Topography, soil types, and wetlands are provided from a variety of existing sources. The assignment of each pixel as habitat for a given species was determined by first estimating the suitability of the pixel as breeding and feeding habitat using a hypothesized species-habitat model and GIS data from a circular neighborhood, centered on the pixel, and then determining the appropriateness of the geometry between breeding and feeding habitats. Habitat assessments can be made by predicting species occurrences for existing landscapes. Impact assessments can be made by predicting species occurrences for simulated landscapes that reflect proposed development, timber harvest, or wetland modifications.