Autumn seems a good time to take stock of some topical parts of the editorial process. Nature journal editors have always encouraged discussion of data before submission. Indeed, we regularly encourage submission to the journal after informal discussion, and we receive some of our best papers this way. All Nature journals accept presubmission inquiries through the online submissions system (http://mts-ncb.nature.com/). We provide feedback on formally submitted 'pre-subs' within a couple of days. Rapid response and ease of preparation probably account for the increasing popularity of this format – currently over 20% of all submissions. However, a negative decision often leads to an attempt to further discuss and explain. Although we fully understand the validity of further discussion, the volume of submissions necessitates that we do not entertain further exchanges. If you feel certain that we have underappreciated your findings, we are happy to read a full submission that allows discussion of all aspects of the work and this undoubtedly makes for a more informed decision.

We are often asked if cover letters are essential. Indeed, it is sometimes evident from a cover letter that an author is expecting us to read no further. This reveals a significant under-appreciation of the thoroughness of the editorial process: Nature journal editors are expert scientists who assess all manuscripts thoroughly. Decisions are made only after significant editorial discussion and, if in doubt, we will consult informally with experts in the field. We save our authors' and referees' time by only reviewing manuscripts that, in our opinion, stand a realistic chance of publication. Cover letters are certainly helpful in framing the author's perception of their dataset, as well as to alert us to related data or individuals they wish to exclude from the reviewing process (we honour exclusions as long as they can be counted on one hand).

It is important to submit manuscripts that are close to our format guidelines, although it is certainly not necessary to fine-tune format requirements on initial submission. The decision whether to publish a dataset in our Article, Letter or Brief Communication format is given very careful editorial consideration, and this will usually be decided firmly only when a manuscripts is deemed publishable in principle.

Given the recent publicity surrounding the editorial retraction of a paper from the journal Cell (see 6 October editorial and The Scientist, September 29th 2005), we wish to restate our policy on refutations. We will consider substantive contributions that claim to refute a major finding published in this journal. We will publish a refutation only if its conclusions are of outstanding interest to our readers and only if the data passes peer review. We usually invite the author of the original paper to comment and we will co-referee this comment (see Nature Cell Biol. 7, 433 (2005)). Substantial errors introduced by the authors or editors are redressed by posting corrigenda or errata, respectively. Addenda may be added at the editors discretion when additional information from the authors alters the interpretation of a paper. If a manuscript is fundamentally flawed or fraudulent, all or some of the authors, or the editors, may decide to retract the paper after due consultation.

We have previously discussed why we value our current system of confidential peer review (Nature Cell Biol. 5, 583 (2003)). Most other journals retain similar systems, although the role of academic editorial boards and the fraction of manuscripts refereed varies. Nevertheless, the debate about systems with more transparency and greater accountability continues.