Above- and belowground biomass of the macrophyte Schoenoplectus maritimus was measured in Camargue (Rhône delta, southern France) using destructive and non-destructive sampling methods. Our aim was to validate whether non-destructive sampling could be used for long-term monitoring of marshes subjected to grazing by cattle and Greylag geese (Anser anser). Height and diameter explained more than 95% of the variation in shoot biomass but the allometric models differed between 2002 and 2003 for the grazed marsh and between the grazed and ungrazed marshes in 2003. This indicates that a generalized model could not be derived and that specific models would have to be established for each marsh. However, we determined that sampling 20 shoots per marsh would be sufficient to establish accurate models. Allometric models underestimated aboveground biomass obtained by destructive sampling and we thus computed correction factors. Total belowground biomass was adequately predicted by the aboveground biomass although the precision of the relationship varied between marshes and years. We concluded that non-destructive sampling can be used to estimate biomass of S. maritimus but that the technique must be adjusted for each study.