We suggest some pointers to guarantee the long-term storage, accessibility and reliability of international research data sets.

Data owners, peer-reviewed journals, research institutes and universities have cited various problems with the feasibility of establishing global databases — including maintenance costs, restrictive data-exchange policies and conflicts of interest.

Research organizations need to start implementing management plans for their own data, as is happening at Wageningen University in the Netherlands. Institutional policies, for example, should mandate that researchers store their data in public repositories.

Setting up distributed databases that are owned by the institutions responsible for the data could also extend data sharing. These institutions would develop their own data policy, and could, for example, pinpoint and resolve any conflicts of interest at an early stage.

Funding agencies and all journals should use their influence to counter restrictive data-sharing policies — for instance, by ensuring that researchers make their data publicly available soon after publication (see, for example, go.nature.com/o5ykhe).

Clear incentives for storing data must replace objections. Collaboration is key to constructing useful databases, both distributed and global.