I will second Ryan in his assertion that post-processing is saved for later. Anything you can fix with an edit is better served with that edit than with noise reduction. It doesn't matter how good the algorithms and processors are, noise-reduction will affect the parts you want to keep as well. Experience helps you figure out when you're reaching that point of diminishing returns. It also helps to know what is going ot be happening fx/music-wise in a given scene, because there may be opportunities for freqeuncy masking that make noise reduction a superfluous endeavor. That all being said I'll give you the run-down of how I use RX2. I haven't used Soundtrack Pro, but there should be some corollaries to how I use things in Pro Tools.
After my edits are done, I duplicate the playlist on every track that needs noise reduction. I try to keep clips grouped together by their source. So, a particular character's dialog, shot on the same day, in one location, with the same mic will be on the same track...or at least a group of tracks next to each other. My goal is to do as little processing as possible, so I start with the most egregious offenders. This is usually hum and steady state noise. I'll start with Denoiser (audiosuite) in Pro Tools to get rid of noise beds, setting the plug-in to process by clip, rather than as an entire selection. Denoiser will frequently take care of a lot of the hum as well (if there is any), so that's one less process. I'll work my way through the audio files, again...only the ones that need it, and process them directly on the timeline. When the files are processed, new clips are created in their place on my newer "duplicate" playlist. If I don't like something that's happened, it's easy to temporarily switch playlists and copy the original file back into my working playlist.
If I find that there is still some steady state noise that needs to be dealt with, I'll touch it again. You're better off using 2 or three passes with mild settings than 1 that's a massive hammer. I do all of this in Pro Tools, because there's not a lot of advantage to doing these "gross" adjustments in stand-alone. Once I'm done with this, I'll make a list of clips that need additional fine tuning to deal with momentary noise issues (mouth noise, car horns, clunky feet, etc.). If any of those files haven't received any processing so far, I'll duplicate the file via "Consolidate Selection." That way, I preserve the original file in case I need to go back to the drawing board. Once I've got my list and am confident that all of the files are safely duplicated (either through the Audio suite processing, or bit-for-bit copying), I'll close out of Pro Tools.
I'll then load up RX2 as stand alone and navigate to the "Audio Files" folder for my project. Using my list, I'll do any necessary spectral repairs directly to the files in that folder. Because I am working on the files that are in my timeline, and haven't changed any file names, all of the retouched clips will be perfectly in place when I re-open Pro Tools. Note that I haven't done any processing to my original audio files, everything on my list should be those previously mentioned duplicates.
After that, I'm usually ready to start mixing. Your final mix is what would go back to your visual edit timeline.