Strategies for dealing with climate change must incorporate and quantify all the relevant uncertainties, and be designed to manage the resulting risks 1 . Here we employ the best available knowledge so far, summarized by the three working groups of the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR5; refs 2-4), to quantify the uncertainty of mitigation costs, climate change dynamics, and economic damage for alternative carbon budgets. We rank climate policies according to di erent decision-making criteria concerning uncertainty, risk aversion and intertemporal preferences. Our findings show that preferences over uncertainties are as important as the choice of the widely discussed time discount factor. Climate policies consistent with limiting warming to 2• C above preindustrial levels are compatible with a subset of decision-making criteria and some model parametrizations, but not with the commonly adopted expected utility framework.Many of the uncertainties surrounding climate change are difficult to quantify and depend on the judgement of experts and on the type of model used to generate future scenarios. Each model produces a distribution over the possible states of nature (for example, cost of mitigation, temperature increase, or economic damage from climate change), and these distributions might differ from model to model. How should we select climate policy in the face of these uncertainties? This paper addresses this question using a framework that accounts for both state uncertainty (for example, the distribution over states of nature) and model uncertainty (for example, the different models (or experts) which generate distributions over states) 5 . We investigate a variety of preferences and assumptions over these two types of uncertainty. A special case is the subjective expected utility 6 framework, traditionally used in economic evaluations. However, an expected utility setting might not work when the information is incomplete and ambiguous, which is clearly the case for climate change 7 . Moreover, people are known to approach risks and uncertainties differently 8 . The proposed setting allows us to explore additional decision-making criteria to deal with uncertainty, in the spirit of refs 7,9. Alternative statistical techniques, consistent with Bayesian approaches, have been developed to cope with model uncertainty 10 . Model weighting is an active topic in climate research 11 , where historical observations provide a basis for model evaluation, although it is not commonly used 12 . Although our framework is sufficiently flexible to accommodate different prior probability measure over the set of possible models, our baseline model assumes a uniform prior with equal model weights.The literature on the role of uncertainty in climate policy making has mostly relied on either analytical or simplified integrated assessment models (IAMs), such as DICE (ref. 13). In such contexts, different decision-making criteria and preferences over risks have been shown to have a sig...