Due to limited sampling of the occurrence of species, Geographical Information Systems (GIS) are being used to model the distributions of organisms to better understand their ecology, to determine current and past distributional patterns of individual species and of biodiversity, and to correlate distributions with climatic variables to predict potential effects of climate change. Modeling the distribution of plants or animals is a multi-step process, and an important primary step is the collection and evaluation of occurrence data, which may be available from a number of types of sources. The butterflies of Ohio, U.S.A. present a unique opportunity to evaluate differences in modeled distributions resulting from the use of data from two different sources of occurrence data. These sources were a natural history collection (NHC) database consisting of over 22,000 records (all before 1992) and a long-term monitoring program (LTMP) database with over 400,000 data entries (1995-2004). Resulting butterfly distribution maps from the NHC database were found to be less robust than the maps resulting from the LTMP database, but not significantly so. Modeled distributions of wetland butterflies, of concern in Ohio due to habitat loss, were found to differ significantly between the two databases, indicating that climate is a poor predictor of the distribution of this group. These and other results illustrate how the unique biases in natural history collections and long-term monitoring programs affect the outcomes of modeled distributions and will assist ecologists in evaluating these two types of data as potential sources of organismal distributional information.