Concern exists about the suitability of laser spectroscopic instruments for the measurement of the 18O/16O and 2H/1H values of liquid samples other than pure water. It is possible to derive erroneous isotope values due to optical interference by certain organic compounds, including some commonly present in ecosystem-derived samples such as leaf or soil waters. Here we investigated the reliability of wavelength-scanned cavity ring-down spectroscopy (CRDS) 18O/16O and 2H/1H measurements from a range of ecosystem-derived waters, through comparison with isotope ratio mass spectrometry (IRMS). We tested the residual of the spectral fit Sr calculated by the CRDS instrument as a means to quantify the difference between the CRDS and IRMS δ-values. There was very good overall agreement between the CRDS and IRMS values for both isotopes, but differences of up to 2.3 ‰ (δ18O values) and 23 ‰ (δ2H values) were observed in leaf water extracts from Citrus limon and Alnus cordata. The Sr statistic successfully detected contaminated samples. Treatment of Citrus leaf water with activated charcoal reduced, but did not eliminate, δ2HCRDS- δ2HIRMS linearly for the tested range of 0–20 % charcoal. The effect of distillation temperature on the degree of contamination was large, particularly for δ2H values but variable, resulting in positive, negative or no correlation with distillation temperature. Sr and δCRDS – δIRMS were highly correlated, in particular for δ2H values, across the range of samples that we tested, indicating the potential to use this relationship to correct the δ-values of contaminated plant water extracts. We also examined the sensitivity of the CRDS system to changes in the temperature of its operating environment. We found that temperature changes ≥4 °C for δ18O values and ≥10 °C for δ2H values resulted in errors larger than the CRDS precision for the respective isotopes and advise the use of such instruments only in sufficiently temperature-stabilised environments. Copyright © 2011 John Wiley & Sons, Ltd.