The establishment of a woody crop component within dryland agricultural systems in Australia is gaining momentum. Perennial woody crops are assumed to prevent recharge to groundwater and thus control landscape-scale salinization. To optimize the design of these new farming systems it is important to (1) understand the factors limiting woody crop growth and (2) be able to predict survival and growth. On a hill slope, in a water-limited site in southwestern Australia, tree-height was measured across several established belts of a woody crop (Eucalyptus polybractea). The major factor limiting growth was soil depth, with tree-height limited by depths at < 4 m. The tallest trees were associated with a soil depth of 4-6 m in the mid slope position of the sandplain landform. To model growth across the field site, tree-height was regressed against geophysical (magnetic, radiometric, electromagnetic) and topographic data using a multivariate regression tree method. Radiometric and topographic data distinguished two tree-height units which were associated with different soil depths. Radiometric data identified the sandplain landform which included the best growth evident at the field site. Within the sandplain landform, shorter trees were associated with lower elevations. This paper demonstrates the utility of radiometric and topographic data as tools for identifying suitable sites for the establishment of woody crops.