According to my experience, I felt it took normally a week or 2 at most before the deadline. No matter how long has the review time a week perhaps enough if there is no big argument and of course it took longer while doing simultaneous tasks :-)
I do not like to extend my review over multiple days. I can read the manuscript and make initial notes in 2-3 hours. After that, it depends on what feedback I need to make, but whenever possible, I do the entire review all the same day. For example, I might spend my day on a Saturday reviewing one paper.
Michael W. MarekThank you very much for sharing your way of reviewing. According to my experience, sometimes I also have done the same way. I also have submitted my review results within a few days after accepting the review request.
it depends of the quality of a reviewed manuscript. If the manuscript of good quality, the reviewing work takes one day. If a manuscript requires more efforts to review, it may take several days, a week, several weeks; depending how busy I am during that period. That's why I do not like to receive a paper for reviewing with a short one-week dead-line. I will not be able to do a good quality reviewing in case if a manuscript requires more attention.
George Stoica Thank you for the comment. It looks like a good method to reduce the review burden. But, sometimes reviewing journal papers may be more time consumed than conference papers.
Roman Croitor Thank you very much for your comment. To make a good review, we have to have enough time. In my case, when I am not very busy, I accept one-week review requests. Otherwise, I also refuse a one-week review.
An interesting question. I typically treat reviews with urgency (just like I would love for my papers) and usually accept only reviews for journals indexed in SCOPUS, WOS or both.
Because most of such reveiews get to me because of my previous PUBLICATIONS in the subject area, 1 week is enough to give constructive comments on a submission.