I've probably done 15 or so over the last 6 months and have settled on a method. My 2 devices drift too much for cutting between tracks to match sources (drift is ~10 ms/minute).
To my ear, once you're out by a few milliseconds, phasey things start happening. It's easiest to hear in the vocals. I tried a bunch of different stretch applications and found the one in Wavelab to sound excellent. It took a little number crunching the first time around but I now know the stretch factor for the 2 devices and it's always the same (either 100.018 or 99.982 depending on which one I'm matching).
In wavelab, go process, timestretch, set to high quality and preserve pitch. Then play with the ratio setting - you can use the other settings but I've discovered that the only accurate predictor of the file length comes from the ratio setting (up to 3 decimals). Because of that, it's not worth trying to set it anywhere else.
I have wavelab set to use 32 bit temp files - this may help the sound quality.
So you can either start by determining the gap between 2 points and calculating your desired stretch, or just zero in by trial and error. If you go the latter way, setting the different stretches up in multitrack is a good way to look at them to guess where to go next.
When I do a matrix I usually have one 16 bit source (soundboard>jb3) and one 24 bit source (mics>laptop). I've stretched each one to match the other. I'm not sure which is best to leave and which should be stretched. You could argue doing it tothe 24 bit source because it may provide more accurate data to work with. At the same time, why touch the better source? In any case, to my ears, the wavelab stretch sounds good.
Then I save it in 24 bit format and load the tracks in Cool Edit Pro multitrack. Line 'em up, adjust amplitude, do any eq (often a bass rolloff) and mixdown to 32 bit. Fades. Save in 32 bit fp. Back to wavelab for UV22HR dithering.
So there are many methods and software. This is one stretch approach. The other more common one is to edit between each track to realign the waves. You delete during the gaps between songs where the edit will not be noticeable. You want to get smooth edits (I know CEP has an auto smoothing feature that will blend smoothly with nearby samples).
Either way, it's guaranteed to be a time consuming process... but the results can be great if you get it right!