Considered endangered by the U.S. Fish and Wildlife Service, the Golden-cheeked Warbler (Setophaga chrysoparia) breeds exclusively in the juniper-oak (Juniperus ashei-Quercus spp.) woodlands of central Texas. Large-scale, spatially explicit models that predict population density as a function of habitat and landscape variables can provide important insight for its management and recovery. We used distance sampling to model detection probability and to estimate the density and abundance of singing male Golden-cheeked Warblers on Fort Hood Military Reservation. We used an information-theoretic approach to evaluate hypotheses concerning the effects of proportion of forest type and forest cover, forest-edge density, and patch size on density. We fitted generalized linear models with detection probability as an offset term to predict density as a function of the habitat and landscape variables, calculate a model-based density and abundance estimate, and map density across the area sampled. The design-based estimates were 0.39 males ha^-1 and 7557 singing males. The most supported model contained proportion of forest type and forest cover, both of which had a positive effect on density, as well as forest-edge density, which had a negative effect. The model-based estimates of 0.39 males ha^-1 and 7571 singing males were greater than estimates extrapolated from intensive territory monitoring. Knowledge of factors affecting Golden-cheeked Warbler density can be used to inform recovery efforts, and our density model can be used to assess the effects of various activities proposed for military training and of environmental disturbance on warbler densities.