I used to do that, too. What I discovered is that there's never enough time. My "stuff to research" file was growing faster than I was making any kind of dent in it, and very little of that stuff was even remotely connected to billable projects.
Here's what I do now:
If an article is directly related to a current project, I treat it as project support materials: file it in the project file, create a next action for followup research like looking up more information online, etc. I take notes about it in my main research notebook, summarize those notes in Evernote, and capture the electronic link to the article in CiteULike (technical journals) or Furl (everything else). Once the project is complete, I file everything away together, and purge on a regular cycle. (More or less yearly.)
If an article is just interesting, but not related to a current project, I do pretty much the same thing, minus the followup Next Actions. Then I recycle the physical paper. The Evernote and Furl/CiteULike summaries let me track the original down should I ever need it again. (I treat interesting online stuff more or less the same, minus the physical paper handling.)
If an article isn't interesting enough for one of these two treatments, it isn't interesting enough to keep.
Print materials that I haven't read yet go into a To Be Read file, which I pull from whenever I have time and interest. When the file exceeds the space I've allocated to it, I go through and throw things in the recycle bin until I have more room.
Katherine