American badgers (Taxidea taxus) are large members of the weasel family Mustelidae. Badgers are important predators and creators of burrows in ecosystems in which they occur, but they are not well studied. Their range occurs over most of North America; however, most studies of badgers have occurred in California and the northern portions of their range, while few have occurred in southern habitats. Badger density has been estimated in Utah, Idaho, Wyoming, and British Columbia with no estimates from desert habitats. To measure badger density in a desert habitat, we placed automatic cameras at anthropogenic water sources (drinkers) in the Chihuahuan Desert, identified individual badgers by their dorsal head stripes, and estimated density using spatially explicit capture-recapture analysis. We modeled the detection process for individuals as a function of time, survey region, a learned and trap-specific behavioral response, and a finite mixture model. We then hypothesized that soil composition, soil depth, and land cover influence variation in badger density. From 1,282 camera-nights, we recorded 301 badger visits (23.5 visits per 100 camera-nights) and identified 30 individuals 170 times. The top model included a trap-specific behavioral effect for λ0, a finite mixture model for the σ parameter, which controls the spatial scale over which the expected number of detections declines, and indicated that soil depth was an important covariate explaining variation in badger density. Estimated badger density was 0.10 badgers/km² (95% CI = 0.03–0.31), which is considerably less than the reported density estimates for nonendangered nondesert populations. Our method helps provide researchers with an ability to estimate density for American badgers in arid ecosystems. With modification, our method may be used across the geographic range of badgers, facilitating better understanding of an understudied species.