There are imo 5 things one has to differ concerning paper managment:

1. Redundancy
   • because of plagiarism in distinct scientific branches there is the possibility you waste time reading the same informatin in different papers several times (bigger problem in humanities than natural sciences)
   • focus on some tools like zotero, citeulike, mendeley, scirate that dont overlap in functions and output and thereby create even more redundancy to you.
2. Staying up to date
   • normally solvable by reading the current articles in a top magazine like nature or science
   • with social research tools like citeulike you can follow single researchers with same or similar research topics and trace what papers they read and share information. If you are beginner in a new branch, following a experienced professional is always a good choice.
   • some scientific branches and online journal sites offer different services, study them. For physics arxiv there is for example the social service scirate for rating newest papers with important infos. Search engines (e.g. ISI - Webofknowledge) for scientific publications sometimes offer distinct alert services.
3. Filter out information you are actively seeking for
   • Nobody will actually do research on exact the same topic/question like you (normal case for a PhD). So some information filtering you still have to do on your own. Otherwise you may miss valuable infos and news.
   • best way to do this currently is in my opition saving your search queries/keywords or simply create an google scholar alert by hitting the "create alert". New papers matching your query will be sent to your mailbox. Save papers you already read in zotero, rate & tag them, so you dont read them twice.
4. Getting aware of information that might be interesting to you (Relevance Paradox)
   • The disadvantage of following other researcher is that your information is basically 2nd hand and average, not every top-notch scientist wants to share his newest secrets or best spotted papers. So of course use the tools listed above, but still keep searching yourself a bit (otherwise any of above social services wouldnt work well - give & get), groups often work on slightly different topics. Most of these social services mainly reduce redundancy and filter out new of old/already known information, but not necessarily highly important & interesting information to you
5. Get information before/as fast as your competitors
   • e.g. Overfly a few very special & up to date small journals on your research topic. Or at least check the rss feeds of those journals for some interesting article titles yourself, if something might be interesting concerning your topic.

Following these rules of thumb you should be able to reduce information overload and still gather important and new infos for your research. When to little to move forward with your research you can still raise your reading time step by step. Often the best secrets & ideas you get anyway from your group mates working on slightly diff. topics, not from papers ;)

These links gives some further good starting points for citation/literature managing

http://libguides.gatech.edu/cat.php?cid=19586
http://libguides.gatech.edu/content.php?pid=130877&sid=1265026

Besides this only few seem to be aware of the search operators AROUND(X) and ~ on google. Tilde(~) in front of a search word also searches for synonymical terms. AROUND(X) searches for sources with only X words between two search terms. I use it especially for finding research papers concerning my own very specific questions/topic. For example:

brain AROUND(4) chimpanzee neuroscience evolution

should give out only sources where brain and chimpanzee are separated by not mor than 4 words. This way you can highly improve the correct context of information you are actually seeking for, while without the AROUND operator you get a lot more papers containing these frequent keywords only in intro/abstract.

I have found this paper about the "three pass approach" very helpful.

A summary:

• First pass (about 10 minutes):
  • carefully read title, abstract and introduction
  • read (sub)section headings
  • read the conclusion
  • glance over the references
• Second pass (about an hour)
  • read more carefully (ignoring details such as proofs)
  • look carefully at figures and diagrams
  • mark relevant references for further reading
• Third pass (up to four or five hours for beginners)
  • virtually re-implement the paper

It might be useful to point out that since a few months Google Scholar now has a feature called "Scholar Updates", listing recent papers relevant to your research. It does so based on your google scholar profile (which is thus a requirement).

The suggested papers are indeed surprisingly well selected.