An invasive variety of Phragmites australis (Poaceae, common reed), the M haplotype, has been implicated in the spread of this species into North American salt marshes that are normally dominated by the salt marsh grass Spartina alterniflora (Poaceae, smooth cordgrass). In some European marshes, on the other hand, Spartina spp. derived from S. alterniflora have spread into brackish P. australis marshes. In both cases, the non-native grass is thought to degrade the habitat value of the marsh for wildlife, and it is important to understand the physiological processes that lead to these species replacements. We compared the growth, salt tolerance, and osmotic adjustment of M haplotype P. australis and S. alterniflora along a salinity gradient in greenhouse experiments. Spartina alterniflora produced new biomass up to 0.6 M NaCl, whereas P. australis did not grow well above 0.2 M NaCl. The greater salt tolerance of S. alterniflora compared with P. australis was due to its ability to use Na⁺ for osmotic adjustment in the shoots. On the other hand, at low salinities P. australis produced more shoots per gram of rhizome tissue than did S. alterniflora. This study illustrates how ecophysiological differences can shift the competitive advantage from one species to another along a stress gradient. Phragmites australis is spreading into North American coastal marshes that are experiencing reduced salinities, while Spartina spp. are spreading into northern European brackish marshes that are experiencing increased salinities as land use patterns change on the two continents.