Resubmitting a paper  (Hopefully this question is not too "soft": I asked it at math.stackexchange.com recently and was advised to ask it here instead).  
I submitted a paper to quite a low-rated journal (impact factor less than 0.3) a while ago, partly because I didn't think that my results were particularly good, but also because at the time I was largely ignorant of the significance of journal rankings. After many months "with editor" it has now finally been sent to a referee, but I'm slightly regretting choosing this journal, as recent feedback has given me the impression that my paper is publishable in a much better journal, and I am at the point in my career (starting to look for my first post-doc position) where I really need some good additions to my publication record.
Now that someone is possibly taking the time to read it, would it be considered bad form to withdraw the paper (or wait until the review comes back, assuming it is favourable), and submit it elsewhere? And is it worth the risk? When considering job applications, how much emphasis do people actually put on where a person's papers have been published, as opposed to how many have been published?
Any advice much appreciated.
 A: A kind of compromise solution: leave your paper where it is, and, given that you now realize that your result is worth more publicity, write a short note (like 1 page or so) where you quote your paper and describe its main results without proofs (possibly, adding small improvements or remarks). A journal expressly aimed to this end is e.g. Comptes Rendus Mathematique, but many other do it as well; your advisor may give you a good hint.
A: "how much emphasis do people actually put on where a person's papers have been published, as opposed to how many have been published?"
Quite a bit!
On the other hand, if you withdraw and resubmit, it is unlikely that your paper will be published by the time you are on the job market, so that might be irrelevant at this point.
A: I agree with the others that it's probably not worth withdrawing the paper. But going forward I recommend that you try to consult peers and mentors about everything you do professionally (what papers to write, where to submit a paper, where to apply for jobs, what kind of talks to give, etc.). Try not to think about and decide things like where to submit an article unilaterally. You don't have to do exactly what anyone says, but you should have a good sense of what other people think.
A: Since you are so early in your career, in my opinion you are much better off leaving the paper where it is and hopefully it will be published soon. It's possible that you could have initially aimed higher, but because the refereeing process can sometimes take 6 months or even a year, you definitely don't want to withdraw it now- it could delay acceptance and publication by a long time. Even if you get a favourable referee's report, there's no guarantee that resubmitting it to a better journal will result in a favourable report.
At your stage, it's better to get papers accepted and published, even if they're at lower-rank journals, than to be waiting over a year for a decision by a better journal. 
A: If you decided to withdraw your paper, you should do it right now and not wait for a report. But think twice, and do it only if you have reasonable hopes to get it published quickly in a good place.
A: Research mathematics is a small world. It is rather likely that the people who are currently editing and refereeing your paper will play a significant role in the development of your career in the next 5 years or even more. They might not take too kindly to the withdrawal, especially if they think that the move was not really warranted (i.e., if they don't think that your result is really much too good for the journal you submitted). So think to yourself: is this a risk worth taking? Is your result that badly served by the current journal?
As for the impact factor, have you really tried to find out what people think about those? Deans and other paper pushers might put a lot of store by them, and so it's definitely something to keep in mind in the long run, but your colleagues' opinion of a journal is not based on a somewhat artificial numerical construct, so for your immediate future, the actual name and reputation of the journal is much more important than a somewhat arbitrary numerical score. (It is a fact that the math journals with the highest impact factor are not necessarily the most prestigious ones.)
A: You probably chose that journal with a good reason (i.e. not at random). I know of many fantastic results in "non-top" journals (e.g. Szemerédi's theorem), so please don't worry. If your results are good, they will be recognized and used. Of course some strategy in life is useful, but personally I don't think it is a good strategy to delay publication and annoy several people (editors, referees) at the same time.
