'Sound science' is one of those phrases repeated so often by politicians in recent years that it has come close to losing all meaning. In endless debates over this or that piece of environmental or health regulation, each side invokes 'sound science' to mean anything that supports its end of the argument.

Now US government agencies are being asked to bring more precision to the term. They will soon be required by law to ensure that the scientific information they disseminate meets standards for “quality, objectivity, utility, and integrity” (see page 249).

The move seems to have started as a manoeuvre by industry-backed groups — which helped to draft the law — to make it harder to enact federal regulations. But it could nevertheless prompt a useful discussion of what constitutes 'quality' research, and what the role of science should be in shaping policy.

In public discussion of guidelines for implementing the new law, scientists have argued, perhaps defensively, that the current system of peer review assures quality control. But peer review isn't infallible (see page 258), and President George W. Bush's White House, which casts a wary eye on federal regulation, calls for an additional standard of 'reproducibility'. By this it means that data and research methods supporting 'influential' government actions should be transparent enough for others to achieve the same result using the same tools.

In a world where everyone acts in good faith to practice 'sound science', that standard might be reasonable. But if industry uses the new law simply to gum up the regulatory process with legal challenges, or keeps sending scientists running back to the lab for one more test before accepting their results, this would be far from reasonable. If that happens, scientists and the wider public should rightly demand another kind of transparency — one in which industry lobbyists are required to come clean about their motives.