We present a method for measuring changes in the atmospheric O2/N2 ratio based on data from a gas chromatograph (GC) equipped with a thermal conductivity detector (TCD). In this method, O2 and N2 in an air sample are separated on a column filled with molecular sieve 5A with H2 carrier gas. Since the separated O2 includes Ar, which has a retention time similar to that of O2, the (O2+Ar)/N2 ratio is actually measured. The change in the measured (O2+Ar)/N2 ratio can be easily converted to that in the O2/N2 ratio with a very small error based on the fact that the atmospheric Ar/N2 ratio is almost constant. The improvements to achieve the high-precision measurement include stabilization of the pressure at the GC column head and at the outlets of the TCD and the sample loop. Additionally, the precision is improved statistically by repeating alternate analyses of sample and a reference gas. The standard deviation of the replicate cycles of reference and sample analyses is about 18 per meg (corresponding to 3.8 parts per million (ppm) O2 in air) . This means that the standard error is about 7 per meg (1.5 ppm O2 in air) for seven cycles of alternate analyses, which takes about 70 min. The response of this method is likely to have a 2% nonlinearity. Ambient air samples are collected under pressure in glass flasks equipped with two stopcocks sealed by Viton O-rings at both ends. Pressure depletion in the flask during the O2/N2 measurement does not cause any detectable change in the O2/N2 ratio, but the O2/N2 ratio in the flask was found to gradually decrease during the storage period. We also present preliminary results from air samples collected at Hateruma Island (latitude 24¿03'N, longitude 123¿49'E) from July 1997 through March 1999. The observed O2/N2 ratios clearly show a seasonal variation, increasing in spring and summer and decreasing in autumn and winter. ¿ 2000 American Geophysical Union