This paper discusses a simple method to determine O-17 isotope excess or deficiency ('mass-independent isotopic composition') in CO2 gas. When applying conventional mass spectrometry of CO2 (m/z 44, 45 and 46) to determine the O-17/O-16 ratio, the C-13/C-12 ratio has to be established separately. This can be achieved by analysing an aliquot of sample CO2 before and after subjecting it to oxygen isotope exchange with a pool of oxygen with 'normal' O-17/O-16 ratio, i.e. with Delta O-17=delta O-17 -0.516 x delta O-18 = 0. Cerium oxide has been shown to be practically well suited for the exchange of CO2 oxygen; the reagent is safe and does not produce any contamination. The CO2-CeO2 exchange reaction has 99.8 +/- 0.7% recovery yield. At 650 degreesC this reaction reaches equilibrium in 30 min and, as tested, results in complete oxygen replacement. A 0 determinations depend on accuracy of CO2 delta measurements: the repeatability of +/- 0.015 parts per thousand (1 sigma) in delta R-45 and delta R-46 determination relative to the working reference results in an error of Delta O-17 as small as +/- 0.33 parts per thousand. Such a precision is sufficient for Delta O-17 determination in stratospheric CO2. The calculated Delta O-17 value systematically depends on absolute R-17 and R-13 ratios in isotopic reference materials, which are presently not yet known with certainty (the 17R value is most important), and may be inadequate for O-17-Correction with a = 0.516. Within the present uncertainty, A 17 0 determined in O-17-enriched CO2 agrees with the value directly measured in the enriched O-2 from which this CO2 was produced. Besides Delta O-17 determination, investigated CO2-CeO2 equilibration may have several other implications. Fast, complete isotopic exchange Of CO2 by reaction with CeO2 may also be employed to get reproducible O-17-correction and, hence, to better monitor small delta C-13 shifts and to isotopically equilibrate mixtures of CO2 gases. Copyright (C) 2001 John Wiley & Sons, Ltd.