THE global budget for sources and sinks of anthropogenic CO2 has been found to be out of balance unless the oceanic sink is supplemented by an additional ’missing sink’, plausibly associated with land biota(1,25). A similar budgeting problem has been found for the Northern Hemisphere alone(2,3), suggesting that northern land biota may be the sought-after sink, although this interpretation is not unique(2-5); to distinguish oceanic and land carbon uptake, the budgets rely variously, and controversially, on ocean models(2,6,7), (CO2)-C-13/(CO2)-C-12 data(2,4,5), sparse oceanic observations of p(CO2) (ref. 3) or C-13/C-12 ratios of dissolved inorganic carbon, (4,5,8) or single-latitude trends in atmospheric O-2 as detected from changes in O-2/N-2 ratio.(9,10). Here we present an extensive O-2/N-2 data set which shows simultaneous trends in O-2/N-2 in both northern and southern hemispheres and allows the O-2/N-2 gradient between the two hemispheres to be quantified. The data are consistent with a budget in which, for the 1991-94 period, the global oceans and the northern land biota each removed the equivalent of approximately 30% of fossil-fuel CO2 emissions, while the tropical land biota as a whole were not a strong source or sink.