Projected current and future potential distribution for several vertebrate species, based on correlative bioclimatic models and projected changes in vegetation biomes. Bioclimatic models were built using the Random Forest algorithm. Projected changes in vegetation were also modeled using the Random Forest algorithm but were produced by Rehfeldt et al. (2012). Projected current distribution is based on the average climate conditions for the years 1961-1990. Projected future distributions are based on average climate conditions for the years 2070-2099 using downscaled (30-second or ~1-kilometer resolution) climate projections from two Global Circulation Models: CGCM3.1 (T47) and UKMO-HadCM3. Both projections use the A2 emissions scenario and are from the CMIP3 (Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report) family of climate simulations. Due to changing climatic conditions, species ranges are expected to shift throughout the course of this century. Modeling subsequent shifts in suitable habitat for animal species, and the resulting changes in species assemblages, represent critical information for resource planners and managers. Developing robust suitability models for large geographic areas can be challenging, in part due to insufficient sampling data and to computational limits associated with modeling large geographies at a fine-grained spatial resolution. To overcome these challenges, I developed a method to model habitat suitability in which I built correlative climate suitability models for 366 terrestrial animal species at a relatively coarse spatial resolution for the entire North American continent using species range maps and 23 bioclimatic variables. I then applied the models to both current and projected future climate data downscaled to a moderately fine resolution for western North America. I refined the resulting climate suitability projections by applying a filter that limited suitability to areas in which suitable biomes were projected to be present. I verified my modeling results using an independent species occurrence data set, finding a median accuracy rate of 70%. I found that incorporating information about biomes into the models resulted in projections of larger climate-driven changes in suitability—on average a difference of about 10%. My results also indicate that study species are more likely to see climate-driven losses than gains in habitat suitability. The percentage of study species projected to undergo a significant net decrease in habitat suitability was double the percentage projected to experience a net increase. These results highlight the shortcomings of many broad-scale models and highlight the need to take finer scale vegetation patterns into account. They also indicate that while many animal species could potentially benefit from climate-change induced increases in habitat suitability, the majority of species may suffer from substantial decreases, complicating future conservation efforts.