Abstract: Distributions of mammalian carnivores are commonly used as indicators of human disturbance and underlie strategies for prioritizing land conservation. The objectives of this dissertation are to examine three factors influencing western North American carnivore distributions and improve the application of species models to conservation planning and land management. I investigated how non-motorized recreation influenced carnivores in urban parks and protected areas in northern California oak woodlands. On average, a greater number of native species occurred in sites that did not permit recreation, and more non-native species occurred in recreation sites. Densities of coyotes (Canis latrans) and bobcats (Lynx rufus) were more than five times greater in non-recreation sites. Differences in domestic dog management policies among recreation sites did not affect carnivores, but species densities were negatively correlated with recreation visitation rates. I also examined how variation in land use intensity and habitat configuration influenced carnivore densities in urban protected areas. Using a model selection approach, I found that increasing intensity of development adjacent to protected areas was an important predictor of decreasing densities of bobcats and gray foxes (Urocyon cinereoargenteus). I used assessor's land use codes to define habitat for the study area and found that densities of coyotes, bobcats and gray foxes were positively associated with the configuration of habitat surrounding protected areas. In Grand Canyon National Park, I compared the effectiveness of four non-invasive survey methods and assessed co-occurrence patterns among nine mammalian carnivores. Transect surveys produced the most detections, camera traps and track plates detected rarer species, and hair traps were generally ineffective. In a null model analysis, two pairs of common mesocamivores—coyotes and bobcats, and coyotes and gray foxes—co-occurred less frequently than would be expected by chance, suggesting that spatial interactions among species occurrences should be considered in the development of predictive distribution models.