Permafrost soils store vast amounts of old carbon which are currently locked under frozen conditions and thus are not contributing to the Arctic carbon flux budget. With projected strongly increased Arctic temperatures by end of this century, the deepening of the active layer will make available a growing portion of this permafrost carbon pool for microbial decomposition, which results in the release of carbon dioxide and methane The eventual increase in atmospheric greenhouse gas concentrations accelerates global warming and describes a feedback loop, the so-called permafrost-carbon feedback. In our modelling study (Schneider von Deimling et al., Biogeosciences 12, 2015) we describe the full cycle of permafrost degradation, soil microbial activity, carbon release and increase in atmospheric greenhouse gas concentrations to determine the strength of the additional temperature increase caused by newly thawed soil carbon. For this purpose, we have developed a simplified, two-dimensional multi-pool model, capturing latitudinal variations of surface climate, and vertical information on soil temperature profiles and carbon distributions. We use multiple lines of recent data sets (such as soil carbon inventories and incubation studies) for tuning our model parameters to observational evidence. The computational efficiency of our model allowed us to run large ensembles over many centuries. By considering differing scenarios of future warming, we describe the full spread of uncertainty in future permafrost degradation and greenhouse gas release inherent to simulations of the permafrost-carbon feedback. Besides modelling the slow process of active layer deepening, we also describe fast thaw in sublake sediments resulting from increased future thermokarst activity. As we account for deep (below 3 meter depth) carbon inventories, we quantify the extent to which permafrost carbon fluxes are enhanced by the contributions from old soil carbon that has been removed from the active carbon cycle for multiple centuries to millennia. In our model description we consider two differing sub-reservoirs of deep soil carbon storage – Yedoma deposits (late Pleistocene ice- and organic-rich silty sediments) and deposits formed in thermokarst lake basins. Both environments differ in soil properties, such as ice content, carbon amount, lability, and age. By describing deposit-specific soil parameters, we simulate differing pathways of future carbon release. We analyse the timing and magnitude of individual carbon dioxide and methane fluxes from these deep carbon stores and discuss their role in enhancing circum-Arctic permafrost carbon release within this century and beyond. Our model simulates permafrost carbon fluxes which strongly increase with global temperatures. Under moderate warming (RCP2.6), we infer cumulated CO2 fluxes from newly thawed permafrost until the year 2100 of 20-58 Pg-C. Under excessive warming (RCP8.5), our model suggests a carbon release of 42–141 Pg. When considering strongly enhanced thermokarst activity in a warmer climate, our simulated methane fluxes proved substantial, causing up to 40 % of total permafrost-affected radiative forcing in the 21st century. The additional global warming through the release from newly thawed permafrost carbon proved only slightly dependent on the pathway of anthropogenic emission in our simulations and reached typical magnitudes of about a tenth of a degree by end of the 21st century. The long-term, permafrost-affected global warming increased further in the 22nd and 23rd century, reaching a maximum of about 0.4°C in the year 2300.