Stable isotope analysis of oxygen (O) is increasingly used to determine the origin of nitrate (NO) and nitrous oxide (N2O) in the environment. The assumption underlying these studies is that the 18O signature of NO and N2O provides information on the different O sources (O2 and H2O) during the production of these compounds by various biochemical pathways. However, exchange of O atoms between H2O and intermediates of the (de)nitrification pathways may change the isotopic signal and thereby bias its interpretation for source determination. Chemical exchange of O between H2O and various nitrogenous oxides has been reported, but the probability and extent of its occurrence in terrestrial ecosystems remain unclear. Biochemical O exchange between H2O and nitrogenous oxides, NO in particular, has been reported for monocultures of many nitrifiers and denitrifiers that are abundant in nature, with exchange rates of up to 100%. Therefore, biochemical O exchange is likely to be important in most soil ecosystems, and should be taken into account in source determination studies. Failing to do so might lead to (i) an overestimation of nitrification as NO source, and (ii) an overestimation of nitrifier denitrification and nitrification-coupled denitrification as N2O production pathways. A method to quantify the rate and controls of biochemical O exchange in ecosystems is needed, and we argue this can only be done reliably with artificially enriched 18O compounds. We conclude that in N source determination studies, the O isotopic signature of especially N2O should only be used with extreme caution. Copyright © 2007 John Wiley & Sons, Ltd.