Linear carbon cycle models, tuned to reproduce the CO2 increase observed at Mauna Loa, independently of their individual assumptions, predict almost identical CO2 concentration trends for fossil energy scenarios assuming a slightly increasing production in the next few decades. The basic information for such prognoses therefore is the airborne fraction observed over the last 20 years. Uncertainties in this quantity are due to possible errors in the estimate of fossil fuel consumption and the corresponding CO2 emission, possible natural fluctuations in the baseline CO2 level, and uncertainties regarding the biospheric CO2 input and uptake as a result of deforestation and reforestation and land management. Depending on different assumptions the effective airborne fraction, defined as the ratio of CO2 increase due to fossil fuel CO2 alone to the integrated CO2 production, might be as low as 0.38 or as high as 0.72, compared to the apparent airborne fraction of 0.55. The effective airborne fraction derived from carbon cycle models, considering only the CO2 uptake by the ocean, lies in the range 0.60--0.70. A value as low as 0.40 seems therefore highly improbable. A high biospheric anthropogenic CO2 input therefore must have been accompanied by a high CO2 fertilization effect. Model considerations, however, are not in contradiction with a high biospheric input with the maximum production before 1958, which also would imply low preindustrial CO2 concentrations in the range 270--280 ppm as reported recently.