Progress in science depends on replication of results, so there is widespread support for the principle that researchers should share the materials needed to reproduce their published claims. Many journals, including Nature, specify this as a condition of publication, and several funding agencies — including the National Institutes of Health (NIH), the Howard Hughes Medical Institute and the Wellcome Trust — expect their researchers to make materials widely available. Even so, failure to share is still common. So we should welcome the NIH's announcement of stronger measures to promote ‘good citizenship’ in the life-science community.

As reported on page 953, the NIH will soon require grant applications to include a specific plan for sharing model organisms and materials that may result from the funded research. Applicants' track records of sharing will also be taken into consideration when their grants are up for renewal. This policy seems to have teeth, but the NIH has so far published only a brief summary statement. The rules will take effect in October, so there is some urgency to clarify exactly how they will work in practice.

Researchers have many motives for refusing to share published materials. Some — such as a desire to maintain a competitive advantage or to use sharing as a way to leverage honorary co-authorship — are contrary to the spirit of publication, and the NIH is right to reject these excuses. But other obstacles to sharing are not so easily dismissed and must still be overcome if the policy is to be effective.

The cost and administrative burden of responding to requests for materials can be substantial, particularly for mice and other animals that are subject to stringent shipping regulations. The NIH has promised to make supplementary funds available to cover these costs, which should help. But academic labs cannot be expected to run animal breeding facilities, so the NIH must also provide strong support for repositories such as the Mutant Mouse Regional Resource Centers, and ensure that they have sufficient capacity to receive and distribute strains in an efficient and timely manner.

Critics of the NIH policy are understandably concerned about intellectual-property rights, but universities are not disinterested observers in the debate. They are entitled to patent their researchers' potentially valuable inventions, and in some cases the imperative to share may clash with the profit motive. In announcing that it will hold not only individual researchers but also their institutions accountable for sharing, the NIH has staked out a strong position. It remains to be seen how it will be implemented.

Refusals to share resources are almost certainly underreported, because researchers whose requests are denied are often reluctant to make a fuss, especially if they believe that nothing can be done. But unless they are willing to report problems, nothing will change.

Journals have a responsibility to pursue complaints, of course, but the real power lies with funding agencies. The NIH has provided a lead; other agencies should now follow to establish consistent and enforceable community standards. These must be embodied in clear policy statements, which must also include contact details for individuals to whom problems can be referred, and clear explanations of how complaints will be handled.