Habitat fragmentation is an issue of major concern for the conservation of biota in landscapes throughout the world. For resource managers to have a sound basis for making decisions, and for conservation biology to mature as a discipline, we must go beyond case-by-case interpretation and be able to predict with confidence the impacts of fragmentation in a range of environments. The reliable forecasting, for example, of those species that will be most vulnerable to fragmentation in a given environment would facilitate a pro-active approach to the management of habitats at the landscape and regional scale. There is a need for workers to formulate hypotheses that generate testable predictions, which in turn are evaluated and further refined by empirical studies. Such hypothesis testing will be most convincing when specific predictions are made prior to the conduct of a study. The purpose of the current paper is to make just such a priori predictions on the vulnerability of woodland birds to local extinction in the fragmented box-ironbark Eucalyptus microcarpa/E. tricarpa forest region of central Victoria, Australia. We present a simple model based on three characteristics of terrestrial species: mobility, population density and habitat specificity. We outline the model and use existing information about 43 species previously recorded in these forests (and about which we have pertinent information) to construct a graded series of expectations concerning their proneness to local extinction. The design of a study to test these predictions also is described.