My first rule is to use articles less than 5yrs...or 10yrs old (unless you are referencing seminal and/or classic literature).
2nd, use maybe your top 25 articles per construct (Is this for dissertation or scientific paper? Even if it is not, you have to impose some arbitrary limits on yourself based on content and the points you are trying to convey.)
3rd use articles that have more than one reference point (i.e., you can reference the intro section, methods, and results; perhaps avoiding some articles that have only one point of reference).
And last, I save pdfs of all my articles (so I can eliminate paper) and use a word document (sometimes a excel spreadsheet) to annotate my articles. So far my dissertation has about 50-75 references, but I'll add more maybe before it's all said and done....but I have about 125 RELEVANT articles saved on my hard-drive, so we all have runaway lit reviews (in private), but I don't know where I picked up the tip to narrow it down to those articles that fully support my writing, and also do realize that when I'm searching & searching for more info...it may be avoidance behaviors that are preventing me doing th actual writing (which is the hard part, right?). Lit reviews are fun, time-consuming and important, but there comes a point when you have to sh-- or get off the pot...ha, know what I'm sayin'?
Good luck, sockit!