RATIONALE The separation and purification of oxygen-argon mixtures are critical in the high-precision analysis of Δ(17) O and δ(O2 /Ar) for geochemical applications. At present, chromatographic methods are used for the separation and purification of oxygen-argon mixtures or pure oxygen, but these methods require the use of high-purity helium as a carrier gas. Considerable interest has been expressed in the development of a helium-free cryogenic separation of oxygen-argon mixtures in natural air samples. METHODS The precise and simplified cryogenic separation of oxygen-argon mixtures from natural air samples presented here was made possible using a single 5A (30/60 mesh) molecular sieve column. The method involves the trapping of eluted gases using molecular sieves at liquid nitrogen temperature, which is associated with isotopic fractionation. We tested the proposed method for the determination of isotopic fractionations during the gas exchange between water and atmospheric air at equilibrium. The dependency of fractionation was studied at different water temperatures and for different methods of equilibration (bubbling and stirring). Isotopic and molecular fractionations during gas desorption from molecular sieves were studied for different amounts and types of molecular sieves. RESULTS Repeated measurements of atmospheric air yielded a reproducibility (±SD) of 0.021 ‰, 0.044 ‰, 15 per meg and 1.9 ‰ for δ(17) O, δ(18) O, Δ(17) O and δ(O2 /Ar) values, respectively. We applied the method to determine equilibrium isotope fractionation during gas exchange between air and water. Consistent δ(18) O and Δ(17) O results were obtained with the latest two studies, whereas there was a significant difference in δ(18) O values between seawater and deionized water. CONCLUSIONS We have revised a helium-free, cryogenic separation of oxygen-argon mixtures in natural air samples for isotopic and molecular ratio analysis. The use of a single 13X (1/8" pellet) molecular sieve yielded the smallest isotopic and molecular fractionations, and this fractionation by molecular sieves can be corrected by the amount of molecular sieve used in the experiment. The reproducibility of the method was tested by the measurement of the oxygen isotope ratios of dissolved oxygen at equilibrium with atmospheric air. We confirmed that the choice of methods for making air-equilibrated water was not related to the magnitude of isotope fractionation, whereas there was a difference between seawater and deionized water.