As for any attempts to measure a researcher's success, it is possible (nay, easy) to criticise citation counts, but I don't propose to do so right now. As it happens, at Dagstuhl I was discussing with colleagues (Artur and Berthold) how it is that some papers get cited more than others. At Monday's meeting, womeone commented that the most important way to get your papers cited is to publish them in highly-respected journals. That's worthwhile, but there are other mechanisms.
- The most important thing to do is to work in a sub-field that is growing. If the community is losing interest in a research topic, you should drop it also. I reckon that all other things being equal, the expected number of citations a paper picks up is exponential in the growth rate of the research field, a theory that has the nice feature that expected citations is non-negative. Note: it is not important that the sub-field be big; it must just be growing.
- Tell a colleague directly "please cite my paper", or, less usefully, "you should have cited my paper". I have occasionally received such a request.
- Write with plenty of co-authors — they may help to publicise the paper by giving talks about it.
- Put the right keywords in the title, or failing that, the abstract, so that it shows up when people search for those keywords.
Here's what not to do: notice that some mediocre papers seem to get cited all the time for no good reason --- somehow they become a sort of virus that infects a body of literature. (Compare with the dreaded "signature virus" that was popular in the Usenet during the early 90s.) There is no point in trying to pursue this phenomenon, because it is completely random.
Added 10.12.07: Submit the paper to a journal editor who is likely to be interested in it, familiar with the topic and hence likely to send it to referees who will also be interested.