Humanity faces a formidable challenge in deciding how to respond to global environmental issues, such as climate change and biodiversity loss. This thesis explores the use of cost-benefit analysis (CBA) in this context. I start in Part I by highlighting the problems with conventional CBA, which include its treatment of the future, and its approach to aggregating costs and benefits across individuals. The rest of the thesis comprises empirical investigations of these two issues. In Part II, I carry out the first rigorous statistical analysis of the economic scenarios developed by the Inter-governmental Panel on Climate Change, which have been criticised as being implausible relative to the historical data. I show that for two of the three regions considered, the criticisms are unfounded, but for the other, the IPCC scenarios are considerably more optimistic than is suggested by the data. In addition, I show that economic growth is poorly described by an exponential function, undermining conventional approaches to discounting. Next, I use data on global and national Ecological Footprints to consider whether environmental limits will prevent future economic growth. I find that previous studies using this data have been biased towards pessimism by ignoring technological progress, and I present two novel analyses that incorporate it. I find no evidence that future economic growth will necessarily result in increasing global Ecological Footprint. In Part III, I carry out a partial CBA of a biodiversity conservation project to protect the Ranomafana-Andringitra forest corridor in south-eastern Madagascar. I begin by investigating the linkages between economic growth, urbanisation, and forest cover change and then develop projections of deforestation and species extinction. I show that the local opportunity costs of conservation are likely to be positively linked to its biological urgency, and hence its global benefits. Although the benefits of conservation strongly outweigh the costs in monetary terms, when corrected for the diminishing marginal utility of income, the net present value of the project is strongly negative. Therefore unless compensation is complete and efficient, conservation will reduce social welfare, due to the extreme poverty in the region. Because complete compensation is difficult to achieve, these results call into question the present emphasis on directing conservation efforts towards low-income countries. The thesis shows that the simplifications inherent in conventional CBA can produce misleading results when applied to complex global environmental issues. In particular, assuming that compensation will be complete and costless could encourage decision-makers to adopt projects that are seriously detrimental to social welfare. CBA cannot be both value-free and decisive. I therefore outline a generalised CBA approach that is capable of incorporating important societal value judgments.