Departments of Transportation (DOTs) are paying more attention to wildlife connectivity needs across highways. There are over 600,000 bridges along US highways, approximately 75% of which are over waterways and could be useful for wildlife connectivity. The purpose of the proposed project was to develop an accessible model that DOTs could use to predict wildlife use of existing culverts and bridges. We collaborated with the DOTs and/or Departments of Fish and Wildlife/Parks (etc.) of AZ, CA, CO, FL, GA, ID, ME, MT, NV, OH, OR, and VA and the Alberta MOT to describe a predictive model for wildlife use of existing structures, based on evidence of use of these structures. Using camera trap data from CA, CO, ME, OR, VA, and WA, we developed a statistical model using structure and near-landscape characteristics, and evidence of wildlife use from camera traps. Our approach relied on four types of data: individual camera trap observations of wild animals, the dimensions of existing structures, the land cover surrounding the camera trap sites, and climatic conditions at the time of observation. Observations were aggregated in two ways: 1) by location for the occurrence or non-occurrence of each animal species and group of interest (“location model” dataset); and 2) by sum of the total number of occurrences of each animal species and group of interest, to find the frequency of structure use for each species (“animal-frequency model” dataset). The two aggregated datasets were used to create two models: 1) determination of whether or not a given species or group will use a given structure (location model) and 2) measurement of how often a given species or group will use a given structure (animal-frequency model). For both models, model development was guided by a combination of Akaike information criterion driven stepwise regression and human-driven variable selection. We developed predictive models for 26 species (e.g., bobcat) and species groups (e.g., deer). We also combined the coefficients for all variables for each species to make two determinations: 1) what structural conditions (dimensions and location) do individual species prefer and 2) given certain structural conditions, what is the likelihood of species or species group use of the structure. Model (1) can be used to inform planning and siting of new wildlife crossing structures that could suit particular species or groups of species. Model (2) can be used to inform assessment of expected wildlife use of a network of existing structures, for example, all bridges and culverts in a state. We are currently developing a website for state DOTs to use the model to predict which structures are likely to be useful for wildlife and which not. This will assist their planning and programming wildlife connectivity enhancements for areas not served by existing structures.