Integrated Assessment Models (IAMs) are increasingly being used to inform policymakers about the likely societal and economic consequences of climate change. An example of such a study is the Letter in this issue by Frances Moore and Delavane Diaz (page 127), which uses a widely applied IAM called DICE to show how the effects of climate change on economic growth affect estimates of the social cost of carbon; see also the accompanying News & Views article by Andries Hof (page 99).

IAMs are potentially powerful tools, and results from them have already been incorporated into IPCC Assessment Reports. But just how generally reliable and informative are the projections emanating from the present generation of models?

A typical IAM combines a simple model of the Earth's climate system with a social science or economic model. IAMs necessarily include assumptions — and simplifications — about how the physical climate system works as well as the interaction of demographic, political and economic variables. It is essential, if IAMs are to effectively help shape climate change policy, that these underlying assumptions as well as model inputs are made explicit. Only then can experts probe the robustness and meaningfulness of their outputs. Just as importantly, the equations driving IAMs need to be specified so as to allow users to understand the mechanisms responsible for the model projections and predictions. Such transparency should help avoid the impression that IAMs are 'black boxes', the inner workings of which are inscrutable and impossible even for experts to assess whether model projections are realistic.

The issues of model robustness and consistency between models have occupied the minds of climate researchers for years. A major aim of the Coupled Model Inter-comparison Project is, for example, to assess and improve the performance and reliability of global coupled ocean–atmosphere general circulation models used to predict future climate under a variety of emissions scenarios. The design and organization of the latest phase of the project — CMIP6 — was finalized at a meeting in October 2014 in Germany; full details should be available in April this year (http://www.wcrp-climate.org/wgcm-cmip/wgcm-cmip6).

IAM researchers are now emulating the efforts of climate researchers by instigating their own model inter-comparison projects (MIPs). This is likely to be a fruitful enterprise. To illustrate the point, in a Review on page 119 of this issue, Massimo Tavoni and colleagues discuss what can be learned from a recent MIP in the context of post-2020 climate negotiations, and actions needed at a regional level to help meet the widely discussed 2 °C target.

We can expect many similar studies to appear over the coming months and years. It will be interesting to see how they fare.