Estimates of the Arctic soil carbon pool have more than doubled in size recently. However, it remains very uncertain how much of this carbon will enter the atmosphere as a result of future global warming. We review in this paper recent estimates of the Arctic soil carbon pool, and experimental and modeling studies. We conclude that the part of the carbon pool that is particularly vulnerable to decomposition at short and long time scales needs better quantification. Current models are insufficiently equipped to quantify the carbon release at rapid thaw of ice-rich permafrost, which also may occur at mean annual air temperatures far below zero. Also the resilience of the carbon sink of ecosystems subject to permafrost thaw needs better quantification. © 2012 Elsevier B.V.