More than 150 journals and many research-funding charities have endorsed the ARRIVE guidelines for reporting research that uses animal models (C. Kilkenny et al. PLoS Biol. 8, e1000412; 2010), but we find that they are being largely ignored. This could undermine data reproducibility and model credibility, and might obstruct translation into human therapy (S. Landis et al. Nature 490, 187–191; 2012).

For example, of 180 papers on multiple sclerosis listed on PubMed in the past 6 months, we found that only 40% used appropriate statistics to compare the effects of gene-knockout or treatment. Appropriate statistics were applied in only 4% of neuroimmunological studies published in the past two years in Nature Publishing Group journals, Science and Cell (details available from D.B. on request). Many journals are therefore failing to ensure that the basics of experimental design and data analysis are respected.

Simply introducing guidelines is not enough. The issue requires greater editorial oversight (perhaps using a tick-box questionnaire at submission), stricter refereeing standards and engagement by learned societies.