The Kyoto Protocol has only just come into force, but its 2012 expiration date already looms large on the horizon. A successor agreement faces some formidable political obstacles. But negotiators should put these to one side for now and ask themselves this: what kind of agreement is best justified by our current knowledge of climate science?

Negotiations on Kyoto II are due to begin at the eleventh meeting of the parties to the United Nations Framework Convention on Climate Change in Montreal in November. Climate science, and the tools that researchers can use to detect global change, have advanced significantly since 1997, when the Kyoto Protocol was negotiated. These improvements must be considered when laying the basis for what comes after 2012.

One important advance has been the study of global carbon cycles — the large-scale carbon fluxes between the biosphere, the atmosphere and the oceans. Siberia, for example, is likely to undergo significant changes in vegetation, wildfire frequency and permafrost abundance, which all have a substantial impact on global carbon transfers (see page 732). Similar changes are being observed in other terrestrial ecosystems, including tropical rainforests and deserts.

Meanwhile, better satellite sensors, in combination with other remote-sensing techniques, provide ever more accurate data about biomass changes on the ground. The Siberian work has demonstrated that it is possible, in principle, to reliably determine carbon flux to and from Earth's main carbon reservoirs.

In a few years' time, a comprehensive, global carbon accounting system, incorporating natural carbon fluxes between the ground, the oceans and the atmosphere, as well as anthropogenic emissions, could be in place. Global carbon accounting, if it can be shown to work, would greatly reduce uncertainty about changes in atmospheric greenhouse-gas levels. And our understanding of climate sensitivity — the response of the climate to changing greenhouse-gas levels — is slowly improving (see, for example, J. M. Murphy et al. Nature 430, 768–772; 200410.1038/nature02771).

All this could strengthen the case for a Kyoto II agreement that would incorporate a target for the parameter of central importance in determining the course of climate change: the concentration of greenhouse gases in the atmosphere. A new agreement could have such a target at its core, advocates suggest. Such a deal would also include emissions targets, but would be flexible enough to allow changes in these secondary targets as more evidence comes to light, or if natural carbon cycles don't behave as expected.

At their summit meeting in March, European Union (EU) leaders announced a goal of keeping global temperature rise below 2 °C over pre-industrial levels. As temperatures have already increased by around 0.7 °C, this would limit future warming to 1.3 °C. The EU has also tentatively suggested that a 15–30% cut in greenhouse-gas emissions should be sought by 2020.

But targets based on temperature or on emission ceilings still suffer credibility problems. With climate sensitivity still under debate, any given temperature target is necessarily associated with a broad range of possible emissions levels. Likewise, any emission target corresponds to a very wide range of temperature-change scenarios.

Focusing on greenhouse-gas concentrations could help policymakers steer clear of this uncertainty trap. An interim target of, say, 400 parts per million of carbon dioxide by 2030 might attract widespread political support. Given advances in global carbon accounting, it would also be more manageable and backed by more solid science than the other options. Such a target would have to be supported by goals for greenhouse-gas emissions, but these could be open to some modification in response to real-world observations and advances in scientific knowledge.

The Intergovernmental Panel on Climate Change (IPCC) — the scientific branch of the convention — is already compiling the latest advances in the science of climate change for its next assessment report, due in 2007. The challenge is to incorporate this knowledge into a viable, fair and effective new agreement.

Given the trenchantly negative approach of the current US administration (which will itself expire in 2008), it falls to the EU and developing countries to ensure that the Montreal meeting makes progress towards drafting a new agreement. Their negotiators, backed by the IPCC's findings and the climate-research community, must be decisive in translating the available scientific knowledge into a workable framework for the treaty that will succeed Kyoto.