Nine months ago we promised to look into trialling double-blind peer review as a choice for authors (Nature Geosci. 5, 585; 2012). We are now pleased to announce implementation: Nature Geoscience and Nature Climate Change will allow authors to choose to remain anonymous to referees, just as referees usually remain anonymous to authors.

But will they indeed be anonymous? Perhaps the most popular argument against a double-blind process is that referees will guess the authors' identities, regardless of whether or not the names are actually sent along with the paper, because the content will give them away. We realise that this may be the case occasionally, and we will rely on participating authors to phrase their paper carefully so as to avoid easy identification. Nevertheless, we are convinced that for many papers the double-blind process will serve to remove unconscious biases.

From our experience, authors who try to guess the identity of a referee are very often wrong. It seems unlikely that referees will be any more successful when guessing the identity of authors. Of course, these hunches are usually not verified or falsified, given our obligation to maintain confidentiality. So there is a widespread (and only too human) overconfidence in the accuracy of one's own assumptions. We strongly recommend that referees who receive a paper within the trial do not ponder who the authors might be; authorship should not, in any case, affect their report.

But even if referees correctly identify the research group that a paper is coming from, they are much less likely to guess who the first author is. One of our motivations for setting up a double-blind trial is the possibility that female authors are subjected to tougher peer review than their male colleagues — a distinct possibility in view of evidence that subtle gender biases affect assessments of competence, appropriate salaries and other aspects of academic life (Proc. Natl Acad. Sci. USA 109, 16474–16479; 2012). If the first author is unknown, this bias will be largely removed.

In a reader survey last year (Nature Geosci. 5, 585; 2012), three-quarters of respondents were supportive of double-blind peer review, with only 16% unconvinced. Interestingly, those who might benefit did not preferentially support a double-blind process: the ratios of males to females, established scientists to young researchers, and people from western countries to scientists elsewhere in the world, were all very similar (down to a per cent or so) between supporters of double-blind peer review and the entire group of respondents.

Determining the impact of a double-blind process on publication outcomes for specific groups of authors will require a larger sample than we will have for a long time. But we will closely monitor uptake of the new double-blind option, as well as the perceptions of participating authors and referees. We are looking forward to finding out if the option to stay anonymous as an author will prove as popular as the reader survey suggested.