The use of a Declining Discount Rate (DDR), in cost-benefit analysis (CBA), compared to the use of a Constant Discount Rate, implies that the policy maker will put relatively more effort to improve social welfare in the far distant future than in the shorter time. The choice between the two discount rates is crucial and linked, for example, to the problem of whether we should fight malaria and AIDS (which have immediate effects) rather than climate change (which is expected to have important long-term effects). In this paper we assess the willingness to pay for (very) distant benefits, which should inform the desirability of policies and projects with immediate costs and distant benefits. DDRs offer an approach to balancing current costs and distant benefits. First we present the existing theoretical justifications for using a DDR, which are mainly driven by the uncertainty of future economic conditions, and show how a theory-consistent optimal trajectory of the DDR can be estimated. For this empirical estimation, we use regime-switching models of the optimal trajectory of the DDR for nine ‘representative’ countries. We then compose a weighted average rate that can be used in CBA of long-term projects that affect the global environment and economy. Finally, we investigate the policy implications of applying this optimal trajectory on the cost-benefit evaluation of carbon mitigation policies and compare our results with those of the Stern Review. This comparison provides empirical evidence that support the major criticism of the Stern Review of assuring high damage numbers by using an arbitrary low and constant discount rate. Our main point in this paper is that when uncertainty is introduced, the case for DDRs and the availability of a reliable empirical method for their estimation become compelling for CBA of long-run policies and projects.