Drought is likely to increase in intensity and frequency across most of the Mediterranean areas due to climate change. There is thus an urgent need to assess differences in the ability of plants to withstand water stress, especially when selecting appropriate species for ecological intervention. This study focuses on Mediterranean-type ecosystems (MTEs) and identifies plant traits associated with drought resistance that are key in differentiating plant functional types in relation to water use. We further discuss how this knowledge can be used by restoration practitioners. The study was conducted in the Banksia woodlands, Southwestern Australia, and six areas across a gradient of water availability were selected. We measured twelve functional traits associated with water use in fifteen plant species. Next, we applied multivariate analyses to examine how traits varied in relation to each other, grouping species based on these traits and investigating similarities within and between functional groups and sites. Functional trait correlations were consistent with the worldwide leaf and wood economic spectra. Among the twelve traits measured, six explained most trait variation: mean xylem vessel diameter (Dave), number of xylem vessels per mm−2 (Ds), leaf mass per area (LMA), stem density (WD), foliar carbon isotope composition (δ13C), and leaf water potential at turgor loss point (πTLP). Species were clustered into five different functional groups. Differences within and between functional groups and sites are reported through their Euclidean distances. Analyses of these traits provided insights into the water-use strategies of native plants, revealing those species with greatest potential to resist water deficits. Such knowledge enables the formation of a more functionally diverse assembly of species bearing complementary traits, which in turn can be used to strengthen resistance to invasion in restored communities. This functional ecological approach is transferable to other and for application by restoration practitioners since the traits selected are relatively easy and cheap to measure and require only simple analytical approaches.