Positive carbon-cycle feedbacks have the potential to reduce natural carbon uptake and accelerate future climate change. In this paper, we introduce a novel approach to incorporating carbon-cycle feedbacks into probabilistic assessments of future warming. Using a coupled climate-carbon model, we show that including carbon-cycle feedbacks leads to large increases in extreme warming probabilities. For example, for a scenario of CO₂ stabilization at 550 ppm, the probability of exceeding 2°C warming at 2100 increased by a factor of between 1.7 and 3, while the probability of exceeding 3°C warming increased from a few percent to as much as 22%. CO₂ fertilization was found to exert little influence on the amount of future warming, since increased carbon uptake was partially offset by fertilization-induced surface albedo changes. The effect of positive carbon-cycle feedbacks on the likelihood of extreme future warming must be incorporated into climate policy-related decision making.