This paper constitutes an overview and synthesis concerning atmospheric and oceanic oxygen and related carbon dioxide, particular attention being paid to potential regulation mechanisms on different time scales. The world atmospheric oxygen reserve is remarkably large, so that a lack of oxygen will not easily occur, whether in confined spaces or in major conurbations. On ''short'' scales, measured in hundreds or thousands of years, feedback processes with regard to oxygen regulation, solely based on atmospheric oxygen variations, would have to be so highly sensitive as to be barely conceivable Only oxygen in the oceans can vary between zero and about twice the saturation - a fact which suggests that here, at least potentially, there exist possibilities for a feedback based on oxygen changes. Atmospheric oxygen production began some 3.2 billion years ago, and has resulted in a net total amount of 5.63 x 10(20) mol (1.8 x 10(22) g) of oxygen, of which 3.75 x 10(19) mol is present as free oxygen in the atmosphere and 3.1 x 10(17) mol as dissolved oxygen in the oceans, the remainder being stored in a large number of oxidized terrestrial and oceanic compounds. For the past 600 million years, atmospheric oxygen has been modelled as ranging between 7 and 30 O-2 vol % (one model even proposes up to 35 O-2 vol %). The present global atmospheric oxygen level remains remarkably constant at a level of 20.946 +/- 0.006 vol %, with a slight decrease of 0.0004 O-2 vol %/yr (4 ppmv/yr) which is counter-correlated to CO2 produced by fossil-fuel and biomass burning. Burning of all terrestrial biomass and humus will potentially reduce oxygen to 20.846 O-2 vol %, while combustion of all known fossil fuel reserves will eventually stabilize oxygen at 20.104 O-2 vol %. The present increase in atmospheric CO2 of about 2 ppmv/yr is linked to a global warming of 0.0125 degrees C/yr, at least in higher latitudes. A similar annual rise in ocean temperature will lead to a release of between 0.2 and 0.5 %/yr of the oxygen present there, and the oceanic oxygen system can be considered as sensitive to global heating. On a small time scale, oxygen is negatively correlated on a 1:1 basis to atmospheric CO2, which is at the level of 0.03 vol % or 300 ppmv (1965 level; 1992 level: 360 ppmv). Thus, close to vegetation, atmospheric oxygen cannot increase by more than this amount (0.03 vol %), and can therefore not exceed 20.976 vol %. The chance of an increase in forest fires due to oxygen Variations is consequently negligible. The atmosphere-ocean exchange budget of CO2 is dominated by the fossil-fuel and biomass fire-derived CO2, of which 39 % is due to CO2 increase in the atmosphere, 30 % is taken up by the oceans, the remainder possibly being recovered by biomass growth. Compared to the atmosphere-ocean flux of CO2, the biogenic ocean fluxes of carbon (''oceanic biological pumps'') are lower by a factor of at least ten (burial of organic matter and foraminifers, and reef growth). Feedback processes of regulation of atmospheric oxygen related to photosynthesis-respiration are difficult to prove for ''short'' periods counted in hundreds or thousands of years, but may be linked to shifts in C-3/C-4 photosynthetic activities both on land and in the sea. C-3 and C-4 photosynthetic pathways react differently to changing environmental conditions such as light, temperature, moisture (on land), CO2 and O-2 levels and nutrients. C-4 photosynthesis is more efficient than that of C-3 with respect to nutrients. At higher temperatures, C-4 photosynthesis is favoured on land and possibly also in the sea. The environmental impact on feedback is discussed in relation to different time scales, and with reference to the earth, its oceans and ocean margins, glacial and interglacial periods, for which the potential contribution by the oceans to the control of atmospheric oxygen changes is evaluated.