Globally, expansion of non-native woody vegetation across floodplains has raised concern of increased evapotranspiration (ET) water loss with consequent reduced river flows and groundwater supplies. Water salvage programs, established to meet water supply demands by removing introduced species, show little documented evidence of program effectiveness. We use two case studies in the USA and Australia to illustrate factors that contribute to water salvage feasibility for a given ecological setting. In the USA, saltcedar (Tamarix spp.) has become widespread on western rivers, with water salvage programs attempted over a 50-year period. Some studies document riparian transpiration or ET reduction after saltcedar removal, but detectable increases in river base flow are not conclusively shown. Furthermore, measurements of riparian vegetation ET in natural settings show saltcedar ET overlaps the range measured for native riparian species, thereby constraining the possibility of water salvage by replacing saltcedar with native vegetation. In Australia, introduced willows (Salix spp.) have become widespread in riparian systems in the Murray–Darling Basin. Although large-scale removal projects have been undertaken, no attempts have been made to quantify increases in base flows. Recent studies of ET indicate that willows growing in permanently inundated stream beds have high transpiration rates, indicating water savings could be achieved from removal. In contrast, native Eucalyptus trees and willows growing on stream banks show similar ET rates with no net water salvage from replacing willows with native trees. We conclude that water salvage feasibility is highly dependent on the ecohydrological setting in which the non-native trees occur. We provide an overview of conditions favorable to water salvage. Copyright © 2011 John Wiley & Sons, Ltd.