The ratio of the stable carbon isotopes, δ13C, contains valuable information on the processes which are operating on the global carbon cycle - climate system. It can help to pinpoint, which sources or sinks significantly alter atmospheric CO2 as δ13C is recorded in ice cores and benthic organisms buried in the sediments. Here we show with the help of the carbon cycle box model BICYCLE how much additional information on carbon cycle and climate dynamics might be extracted from δ13C and where we find significant limitations. Our (paleo-) time frame of interest is spanning from the variability during fast climate fluctuations of the Dansgaard/Oeschger (D/O) events to the rise in the glacial/interglacial amplitudes and shift in the frequency spectra from 40 kyr to 100 kyr during the Mid Pleistocene Transition (MPT).Our findings include:(1) In our model output δ13C of atmospheric CO2 during D/O events is affected largest by the terrestrial carbon cycle, however, the δ13CO2 signal is significantly diluted (faster than CO2 itself) due the gas exchange with the surface ocean if stadial/interstadial transitions are a lot slower than one century.(2) A simulated atmospheric δ13CO2 record over the last 800 kyr (not yet confirmed by ice core records due to a lack of measurements) does not contain any significant power in the 100 kyr periods due to opposing effects of the terrestrial biosphere and the different marine carbon pumps.(3) The rise in glacial/interglacial amplitudes in deep Pacific δ13C by ~40% over the MPT can only be explained if another major process besides those recorded in the LR04 benthic global δ18O stack was in operation leading possibly to a decoupling of SST and vertical water fluxes in the Southern Ocean prior to the MPT (Southern Ocean Decoupling Hypothesis).Limitations:(1) The amplitude of δ13CO2 as recorded in ice cores heavily depends on the gas age distribution in the firn before gas enclosure and leads to a dampening of recorded peaks (in case of EPICA Dome C of up to ~60% for LGM conditions). This makes a knowledge of the pure atmospheric signal (which is the one of interest for a comparison with carbon cycle model simulations) during fast D/O events difficult.(2) The prominent 400-500 kyr cycle found in all deep ocean δ13C reconstructions and the complete lack of this frequency in δ18O still holds some surprises in the understanding of the carbon cycle - climate interaction.Altogether we conclude that δ13CO2 to be measured in ice cores might contain valueable information on past changes in especially the terrestrial carbon storage, but the gas age distribution in ice cores of low accumulation site blurs the original atmospheric signal. Information in deep Pacific δ13C add substanially new pieces of evidence on the climate evolution over the last 2 million years which should be used as much as possible in the interpretation of climate change.