Stable isotope analysis of O2 is a valuable tool to identify O2 -consuming processes in the environment; however, reference materials for O2 isotope analysis are lacking. Consequently, a one-point calibration with O2 from ambient air is often applied, which can lead to substantial measurement uncertainties. Our goals were to develop a simple multipoint isotope-ratio calibration approach and to determine measurement errors of δ18 O and δ17 O values of O2 associated with a one-point calibration.We produced O2 photosynthetically with extracted spinach thylakoids from source waters with δ18 O values of -56‰ to +95‰ and δ17 O values of -30‰ to +46‰. Photosynthesis was chosen because this process does not cause isotopic fractionation, so that the O isotopic composition of the produced O2 will be identical to that of the source water. The δ18 O and δ17 O values of the produced O2 were measured by gas chromatography coupled with isotope-ratio mass spectrometry (GC/IRMS), applying a common one-point calibration.Linear regressions between δ18 O or δ17 O values of the produced O2 and those of the corresponding source waters resulted in slopes of 0.99 ± 0.01 and 0.92 ± 0.10, respectively. In the tested δ range, a one-point calibration thus introduced maximum errors of 0.8‰ and 3.3‰ for δ18 O and δ17 O, respectively. Triple oxygen isotopic measurements of O2 during consumption by Fe2+ resulted in a δ18 O-δ17 O relationship (λ) of 0.49 ± 0.01 without δ scale correction, slightly lower than expected for mass-dependent O isotopic fractionation.No significant bias is introduced on the δ18 O scale when applying a one-point calibration with O2 from ambient air during O2 isotope analysis. Both O2 formation and consumption experiments, however, indicate a δ17 O scale compression. Consequently, δ17 O values cannot be measured accurately by GC/IRMS with a one-point calibration without determining the δ17 O scale correction factor, e.g. with the O2 formation experiments described here.