It is well known that N2 in the ion source of a mass spectrometer interferes with the CO background during the δ18O measurement of carbon monoxide. A similar problem arises with the high-temperature conversion (HTC) analysis of nitrogenous O-bearing samples (e.g. nitrates and keratins) to CO for δ18O measurement, where the sample introduces a significant N2 peak before the CO peak, making determination of accurate oxygen isotope ratios difficult. Although using a gas chromatography (GC) column longer than that commonly provided by manufacturers (0.6 m) can improve the efficiency of separation of CO and N2 and using a valve to divert nitrogen and prevent it from entering the ion source of a mass spectrometer improved measurement results, biased δ18O values could still be obtained. A careful evaluation of the performance of the GC separation column was carried out. With optimal GC columns, the δ18O reproducibility of human hair keratins and other keratin materials was better than ±0.15 ‰ (n = 5; for the internal analytical reproducibility), and better than ±0.10 ‰ (n = 4; for the external analytical reproducibility). Copyright © 2011 John Wiley & Sons, Ltd.