I have a manuscript sat on my desk that I have to review for a journal. It's something I've done a good few times now, but it's surprising how little feedback one gets on the process. I only really get an idea of what's expected from the reports that I get on my own work, which vary a lot. Well, thankfully Thom Brooks is out to put that right. He's working on a piece of advice for refereeing papers (following his previous advice for publishing) and currently inviting suggestions here.