Predicting species occurrence based upon landscape-scale characteristics is a fundamental goal of ecology and conservation biology. Accurately predicting the potential occurrence of a species is fundamental to management activities that involve large areas where sampling is difficult due to logistical or financial constraints. During the summers of 2001-2003 mist nets were used to capture bats in Whiskeytown National Recreation Area (WNRA), Lassen Volcanic National Park (LVNP), and Lassen National Forest (LNF) in northern California. I used logistic regression and Akaike's Information Criterion (AIQ to model species distributions. Models developed a priori were used to determine which variables best discriminated between capture sites and non-capture sites. Prediction models were mapped using Geographic Information Systems. In WNRA, for all bat species combined total edge was most parsimonious, whereas in LVNP elevation was best for all species. Elevation and tree size were important in predicting the occurrence of pallid bats (Antrozous pallidus), in LNF. Results of this study are important to wildlife managers within the study areas because the models can be used to minimize deleterious impacts on bats. Moreover, distribution maps are valuable to bat conservation efforts because they provide baseline data important for evaluating and predicting population responses to management activities. / Department of Biology