Improvements made to an established mass spectrometric method for measuring changes in atmospheric O2/N2 are described. With the improvements in sample handling and analysis, sample throughput and analytical precision have both increased. Aliquots from duplicate flasks are repeatedly measured over a period of 2 weeks, with an overall standard error in each flask of 3–4 per meg, corresponding to 0.6–0.8 ppm O2 in air. Records of changes in O2/N2 from six global sampling stations (Barrow, American Samoa, Cape Grim, Amsterdam Island, Macquarie Island, and Syowa Station) are presented. Combined with measurements of CO2 from the same sample flasks, land and ocean carbon uptake were calculated from the three sampling stations with the longest records (Barrow, Samoa, and Cape Grim). From 1994–2002, We find the average CO2 uptake by the ocean and the land biosphere was 1.7 ± 0.5 and 1.0 ± 0.6 GtC yr−1 respectively; these numbers include a correction of 0.3 Gt C yr−1 due to secular outgassing of ocean O2. Interannual variability calculated from these data shows a strong land carbon source associated with the 1997–1998 El Niño event, supporting many previous studies indicating that high atmospheric growth rates observed during most El Niño events reflect diminished land uptake. Calculations of interannual variability in land and ocean uptake are probably confounded by non-zero annual air sea fluxes of O2. The origin of these fluxes is not yet understood.