How to select a journal? What are good criteria for selecting a journal to submit a paper to?
One criterion per answer, please.
It is easy to group journals by subject and prestige, but is there a thought-process that you use to determine which journal is good for your specific paper?
For example, if you have a clearly top paper, it is not hard to find a journal that will accept it (the only question is which top journal to submit it to), but do you have favorite journals for specific types of projects, papers that are not earth-shattering, or other categories of papers?
 A: There are various issues related to the internet. The author of the paper might want to keep her copyright. I think most journals allow posting of papers on arXive now so posting there even after publication is not an issue in most cases. In fact I have heard of some journals posting their articles on arXive. The manner of submission can involve the internet most journals now have internet submissions systems and that could be an issue. I prefer an internet interface over snail mail but some people might have problems with the internet submission process. 
A: Norwegian "level".
Some official body in Norway assigns journals a "level": 2 is very good, 1 is ordinary, and there are journals that don't get a level at all.  There are about 100 at level 2 in mathematics.  There's some system, which I don't know all of the details about (Harald, are you there?), whereby my department benefits if I publish in higher levels than lower levels.  So I try to publish in higher ranked journals than lower ones.
A: MSC.
I look back at the last N articles in a journal (where N is of the order of a hundred or so) and extract the MSC information.  I then divide that into percentages and according to major and minor modes (ie 53 is major, 53C is minor, I don't distinguish further down the tree).  I then try to see if there's a good match for my paper.
A: See also a previous question: Which are the best mathematics journals?.
A: I guess I should mention eigenfactor as well.
As far as I'm concerned, I don't give this much weight.  It may be a "better" version of impact factor/citation index but as it isn't used (to my knowledge) by those who I'm hoping to influence in that way, there's no point in using it.  And while it may be better, it's still a crude "impact factor" measurement which I completely disagree with.
A: Finally, I guess I should mention how I collate all of that information.  I shan't put the whole details here but anyone interested can see it all at this page on my website.  Unfortunately, the relevant copyright statements (AMS being surprisingly worse than ISI on this one) mean that I cannot redistribute my data and so make it easy for anyone else to use.  However, I can at least make the technique available and that is what is available on the above-linked page.
A: It's perhaps not in the spirit of Noble Academe, but increasingly I find myself looking at the editorial board for the journal. If there are people on it with whom I have actual personal professional contact (e.g. discussion at conference, or a prior exchange of work ideas) then -- other things being equal -- I'll be more likely to submit there.
(This is just for those of us whose primary concern is to aim for a decent journal but avoid the immediate "bin-as-intray". If your work is good enough, this shouldn't matter as much.)
A: One of the standard heuristics I've heard is to look through the journals that the papers your paper cites were published in, and pick one of those.   
This was also discussed in this March 2009 post at Secret Blogging Seminar.
A: I'd like to mention two (quite different) factors: the time it takes the journal to accept the paper is a serious consideration: however outstanding your papers, it's probably useful for it to be accepted in famous journal X instead of submitted to famous journal X when you go on the job market. I think the AMS periodically publishes average processing time for journals in the Notices.
Another factor I've come to try and consider is how expensive the journal is: journals like Inventiones are extortionately expensive compared with AMS journals (including JAMS, to give a journal of comparable prestige). Rob Kirby has information about this on his webpage, and probably there are more recent sources too. Of course I think this criterion is easier to use when you're more established than when you're starting out. 
A: I combine a number of factors.  By the "community wiki rules" I ought to put them one-per-post.
Journal citation index.  There are two versions of this: one by ISI and one by the AMS.  I use both.
Why do I use them?  Because my main reason for selecting a journal is that it gives an indication of the relative importance of the paper to someone who isn't going to read or be able to understand the paper itself.  I fully expect a mathematician working in roughly the same area as myself to be able to make a reasonable assessment of my work without paying any attention whatsoever to the journal it appears in.  However, from time to time someone completely incapable of doing that needs to get a vague idea of how well I'm doing.  At the moment, one of the measures that they use is citation indexes so there is a strong incentive for me to get my papers into journals with the best citation indexes.
I know that citation indexes are unpopular so this answer may well end up at the bottom of the heap!  However, I would like to challenge anyone who votes this answer down to put up a replacement that achieves the same end.
A: There's one piece of information that I would really like to know for each journal but which, as far as I can see, is not (yet) made available:
Time to rejection.
Many journals publish details of the breakdown between submission and publication, but as far as CV points are concerned then once it's accepted then I don't care how long it takes to get actually published.  But if, for the sake of argument, we accept that I want to get my article into a high-ranked journal but have time constraints (for example, if I'm looking for a job) then "time to rejection" is the most important piece of information that I can think of.  Knowing that, I would submit first to a slightly higher ranking journal than I really thought it should go to knowing that I had enough time to get it into a "safe" bet if they rejected it.
A: Many shall find this criterion worthless, but personally I can't help taking the production quality of the journal into account (even if it is not of primary importance). 
Quality paper and ink improve the reading comfort, and some journals (some of Elsevier particularly) have pages with pale ink, or thin paper that let the verso appear. On the other end of the scope are journals like Acta mathematica of Publications mathématiques de l'IHES with nice paper and fonts.
Another aspect of this criterion is the processing quality. To compare two very different experiences, in my first article incredible mistakes have been added after the proofs (expressions like $n/2$ replaced by $n^2$) while the AMS journals do an amazing job, showing you exactly what they changed in your paper (corrected spellings e.g.) and sometimes asking for confirmation. This part you cannot judge before being published in the journal, or discussing with colleagues.
A: "Honest Joe" factor.
(I'm lumping a few together here as this is rapidly going to get out of hand!)  How close does the journal fit with my view of how a journal should work?  Can I retain copyright?  Are my articles free to view for all (rant suppressed with extreme difficulty)?  Is the price reasonable?  Is the publisher on my "good" list?
Sources: http://www.journalprices.com and the AMS journal survey.
A: (This advice isn't my own idea, I am just passing along advice from senior mathematicians', I think I heard it from Bill Fulton.)
Before submitting to journal X, go to the library and look up journal X, and browse around. Would your article fit in there? (Do they publish articles in your field? Do they publish articles that are as technical as yours? Do they publish 95-page articles, such as yours? Etc.)
