More than a million people in Europe signed a petition earlier this year to halt research with animals. One reason why Nature and many scientists are able to defend these experiments is that all involved do everything they can to minimize pain and suffering. Animal experiments are approved only after thorough discussion and are carried out according to strict regulatory controls. Society sees the benefits of animal research, but it does not seek them at any cost.

When breaches of the strict rules that govern animal research occur, it is vital — to both supporters and opponents — that they are investigated thoroughly, and that lessons are learnt and shared. This week, Nature publishes a correction on its website that details such a breach of experimental protocol in a previously published paper (L. Rajet al.Naturehttp://dx.doi.org/10.1038/nature15370;2015).

The relevant experiments grew tumours in mice as a way to test possible treatments. This type of study is common, as is the way they are approved and regulated. Researchers typically plan the experiments and then submit details to an institutional review board for approval. In making its decision, the board follows guidelines set out by a separate body charged with oversight of animal procedures — an institutional animal care and use committee. These guidelines are country-specific, and in the case of tumour experiments should include limits on the maximum tumour size allowed, and instructions to the researchers to monitor both tumour size and signs of distress.

In this case, prompted by a complaint from a reader and following consultation with the authors and the relevant bodies, Nature has established that the scientists did not carry out the required monitoring properly. As a result, some of the tumours grew larger than permitted. These mice could therefore have experienced more pain and suffering than originally allowed for.

As well as writing to correct their paper to mark the breach of animal-welfare guidelines, the authors apologize for the breach. They are right to do so. Cases such as this could provoke a justifiable backlash against animal research. All involved — scientists, institutions, funders and journals — must do more to ensure that regulations are strictly observed.

Nature’s policy is that the corresponding author on a paper that reports experiments with animals must confirm that the research was carried out in accordance with the relevant rules (see go.nature.com/a9pjym). As a result of this case, we are increasing the amount of information we request from authors. In experiments in which tumours are grown, we now require authors to include the maximal tumour size permitted by the institutional animal-use committee, and to state that this was not exceeded. Authors must also provide the source data for any figures that analyse tumour growth.

Nature does not want to publish the results of experiments that have not been performed under ethical guidelines. As such, the authors in this case are correcting their paper to withdraw the portion of the data collected in experiments that the institutional committee concluded were in breach. The scientific conclusions of the paper remain valid and useful, and still stand.

Institutions should do more to make sure that the guidelines they set are respected. At the very least, on completion of each project — and before data are submitted — institutions should verify that approved protocols were followed. Funders and institutions must consider better training for young researchers doing work with animals. And the broader community should continue to scrutinize and improve how it carries out these types of experiment. Discussions are already under way, for example, on whether the control arms of similar cancer studies truly need to let (untreated) tumours grow as large as they currently do. Nature is happy to join these discussions and to help to improve practice.