Typha angustifolia, Juncus maritimus, and Eleocharis palustris were evaluated for de-icing salt removal from runoff water. Plants were exposed to a range of de-icing salt levels (0.2, 0.7, 4, 8, and 13 dS m(-1)) in laboratory-scale subsurface constructed wetlands (CWs) for 2 months under greenhouse conditions. Effluent characteristics, plant height, biomass, and Cl and Na removal rates and uptake were monitored. More water volume was retained in CWs of T. angustifolia (∼60 %) than of J. maritimus and E. palustris (∼37.5 %), which accounted for the electrical conductivity increase in effluents (1.3-1.9-fold). Based on the NaCl removal rate, T. angustifolia showed the greatest phytodesalinization ability (31-60 %) with the highest removal at the lowest salt levels (0.2-0.7 dS m(-1)), followed by J. maritimus (22-36 %) without differences in removal among levels, and E. palustris (3-26 %) presenting a removal rate highly decreased with increasing salt levels. Plant height and biomass were stimulated at low de-icing salt levels, but, at higher levels, T. angustifolia and E. palustris growth was inhibited (tolerance index ∼67 and 10 %, respectively, in the worst cases). Salt amounts in aboveground biomass in g m(-2) differed among levels and ranged as follows: 13.6-29.1 (Cl), 4.2-9.3 (Na; T. angustifolia); 7.0-12.0 (Cl), 2.7-6.4 (Na; J. maritimus); and 0.9-7.6 (Cl), 0.3-1.6 (Na; E. palustris). Chloride and Na translocation decreased with de-icing salt increase in T. angustifolia, while no significant differences were found in J. maritimus, which is interesting for harvesting purposes.