Abstract: There is considerable interest in the use of δ¹⁸O and δD of leaf (δL), xylem (δX) and soil water (δS) as important tracers in analyzing the role of terrestrial biosphere in eco-hydrological cycle. Isotope ratio infrared spectroscopy (IRIS) has emerged as a faster, more cost-effective and field-deployable method for water stable isotope analysis. However, previous studies have also demonstrated the potential for large errors in IRIS. These errors included considerable deviation from isotope ratio mass spectrometry (IRMS) values in the range of (2.64±0.43)‰ for δ¹⁸O and (3.6±0.8)‰ for δD when water stable isotopes were cryogenically extracted from plants/soils with methanol/ethanol-based organic contaminants. As this study focused mainly on establishing correction methods, deionized water spiked with varying amounts of methanol (MeOH, 10~800 μL·L^-1) and ethanol (EtOH, 2~40 mL·L^-1) was used to create correction curves for δ¹⁸O and δD associated with metrics of narrow-band (MeOH, NB) contamination and broad-band (EtOH, BB) determined using Spectral Contamination Identifier (of Los Gatos Research Inc. software, United States). The results showed that while no significant time drift (P > 0.01) existed in correction curves of liquid water δ¹⁸O and δD analyzer (Los Gatos Research Inc.), it varied significantly from other analyzers (P < 0.01). The lightly contaminated (NB < 4 000 and BB < 1.2) δL and δX of winter wheat (Triticum aestivum L.) and summer maize (Zea mays L.) were accurately corrected. The mean differences in isotope ratios between corrected IRIS and measured IRMS were ( 0.11±0.12)‰ for δ¹⁸O and ( 0.7±0.4)‰ for δD. The study discussed the necessity for further validation of the established correction methods for highly contaminated samples (NB > 4 000 or BB > 1.2). It recommended cross-validation between corrected IRIS and measured IRMS using random investigation of small amounts of each unknown species of plants or soils.