Landscape-scale habitat and land use influences on Asian Houbara Chlamydotis macqueenii (IUCN Vulnerable) remain unstudied, while estimating numbers of this cryptic, low-density, over-hunted species is challenging. In spring 2013, male houbara were recorded at 231 point counts, conducted twice, across a gradient of sheep density and shrub assemblages within 14,300 km 2 of the Kyzylkum Desert, Uzbekistan. Four sets of models related male abundance to: (1) vegetation structure (shrub height and substrate); (2) shrub assemblage; (3) shrub species composition (multidimensional scaling); (4) remote-sensed derived land cover (GLOBCOVER, 4 variables). Each set also incorporated measures of landscape rugosity and sheep density. For each set, multi-model inference was applied to generalised linear mixed models of visit-specific counts that included important detectability covariates and point ID as random effects. Vegetation structure received strongest support, followed by shrub species composition and shrub assemblage, with weakest support for the GLOBCOVER model set. Male houbara numbers were greater with lower mean shrub height, more gravel and flatter surfaces, but were unaffected by sheep density. Male density (mean 0.14 km -2 ) estimated by distance analysis differed substantially among shrub assemblages, being highest in vegetation dominated by Salsola rigida, high in areas of S. arbuscula and Astragalus, respectively, lower in Artemisia and lowest in Calligonum. The study area was estimated to hold 1824 males (CI 1645(CI -2030. The spatial distribution of relative male houbara abundance, predicted from vegetation structure models, had the strongest correspondence with observed numbers in both model calibration and the subsequent year's data. We found no effect of pastoralism on male distribution, but potential effects on nesting females are unknown. Density differences among shrub communities suggest extrapolation to estimate country-or range-wide population size must take into account vegetation composition.