There are two readerships of a peer review report: the author(s) of the article being assessed, and the editor making the decision. It reassures me if a reviewer can summarise the most important message(s) of a paper in a brief paragraph at the beginning of the report; I then have the impression that the reviewer can see the larger picture, and has not concentrated exclusively on second order details. As editors cannot be experts in the variety of fields they encounter, such an introductory paragraph can also be helpful to direct the editor's attention to a crucial aspect of the paper that might – because of the author's familiarity with that aspect – have been taken rather too much for granted. If a reviewer, for no fault of her/his own, finds it too hard to summarise the paper, then that should also be noted in the report. As an author, I would feel reassured if a reviewer correctly identified the crux of my thesis, hence indicating that a peer in the field has understood my thesis to the extent of being able to criticise it properly. Particularly in the ‘integrative’ review literature – i.e. papers that seek to present a synthesis that is more than the sum of the parts – it is very important that the new insight is unambiguously and consistently communicated at various levels of the paper: title, abstract, main text and conclusions. That is not a given, and peer reviewers should look out for inconsistencies in high-level messages. Note the merits (if any) of the work in a specific way. Be highly specific, and give reasons, with your criticisms: vague criticisms, particularly if strongly worded, are unlikely to produce a constructive response. Without doing the intellectual work on behalf of the author(s), give suggestions for improvement. Write from your personal expertise base, and with a personal voice: you, personally, are judging the paper, not your field. Unlike the exchanges at the end of a scientific talk, peer review feedback is ‘detached’, and – in the event that the paper is not seen again by the reviewer, or the author does not rebut the reviewer criticisms – unidirectional. Those aspects probably lead to quite a few ‘overly destructive’ reports tainted with wording verging on the offensive. If you truly wish the paper to improve, strong messages should be neutrally worded. Bear in mind that a very strongly worded review report makes it more likely that the author(s) will identify you; that might be intentional, but peer review is not a forum for the continuation of a long-standing personal battle. Neither is it a forum for trivial positive feedback, particularly if overdue: a belated two-liner merely praising the paper and strongly recommending its publication usually just delays the peer review process until another, more substantial, report can be obtained. Should you volunteer to view the paper again? My feeling is that unless you have recommended ‘reject’, the more severe your criticisms, the greater should be your preparedness to re-assess the work. A highly critical review report recommending ‘major revision’, but lacking willingness to re-assess the manuscript, comes across as a veiled ‘reject’ – unqualified unwillingness being a sign of rejection. If you genuinely cannot re-assess the manuscript, briefly state why. Seen from the author's perspective: if taken seriously, a hefty report will cause major investment of time in ‘improving’ the paper. But it might be impossible for editorial, in a further round of peer review, to obtain the same expertise needed to assess the improvements. Finally to language: here only broad assessment is required. Unless the writing is so bad that it obscures the science, you need not go into the detail of grammar/syntax/spelling/punctuation; that is Editorial's responsibility. Furthermore, if you have criticised the scientific aspects of the paper in great depth, adding a detailed critique of the language might dilute the efforts of the author(s) to address other, more important, failings. Make sure too that your own language is unambiguous and concise: your report will have most impact if written in your best English. Andrew Moore Editor-in-Chief