The use of isotope ratio infrared spectroscopy (IRIS) for the stable hydrogen and oxygen isotopeanalysis of water is increasing. While IRIS has many advantages over traditional isotope ratio massspectrometry (IRMS), it may also be prone to errors that do not impact upon IRMS analyses. Ofparticular concern is the potential for contaminants in the water sample to interfere with thespectroscopy, thus leading to erroneous stable isotope data. Water extracted from plant and soilsamples may often contain organic contaminants. The extent to which contaminants may interferewith IRIS and thus impact upon data quality is presently unknown. We tested the performance ofIRIS relative to IRMS for water extracted from 11 plant species and one organic soil horizon. IRISdeviated considerably from IRMS for over half of the samples tested, with deviations as large as 46%(d2H) and 15.4% (d18O) being measured. This effect was reduced somewhat by using activatedcharcoal to remove organics from the water; however, deviations as large as 35% (d2H) and 11.8%(d18O) were still measured for these cleaned samples. Interestingly, the use of activated charcoal toclean water samples had less effect than previously thought for IRMS analyses. Our data show thatextreme caution is required when using IRIS to analyse water samples that may contain organiccontaminants. We suggest that the development of new cleaning techniques for removing organiccontaminants together with instrument-based software to flag potentially problematic samples arenecessary to ensure accurate plant and soil water analyses using IRIS. Copyright # 2010 John Wiley& Sons, Ltd.