How do you use Audio Snap?

What I usually do is create a separate take lane within the track, create a copy and mute the original while bouncing the version processed by AudioSnap. The results typically depend on the amplitude profile of the element. If it's not percussive then I might have to bounce and undo a couple of times to achieve the results I'm looking for. In any case, keeping that muted copy on a separate take lane can help.

I must admit, I've never got my head around Audiosnap... I have played with manually tweaking transient positions to fix timing, but that's it.
 
And now I've taken advantage of the Melodyne offer that comes with SPlat, so I'll be using that from now on, and it's real fancy seeing a polyphonic guitar track in that editing screen 

I do not use AS often, so depending on what you are attempting, it can become complex quickly. I often perform simple tasks, and a couple things to bear in mind is to bounce afterwards, since the rendering is "dumbed down" to minimize CPU load (but can also be a good indicator that you are stressing the audio too much). Also, try to use the fewest transients required when moving (adjusting threshold), and remember you can lasso transients as well. For things in time with each other, you can lasso a set, and click-drag (or ctrl-click drag) them to only adjust what is necessary. Trying to edit every detail always seems to make a mess for me.
 
There is also nothing wrong with cutting a section from a clip and only working on that (bounce it before/after AS) and then reassembling the entire clip (if desired) afterwards.
 
Definitely save before playing with AS though as mentioned above!

When I'm dragging transients around, I leave snap off and just drag stuff closer to the correct position. I do the same with Melodyne too 

I use a Audio Snap manually to tighten up the timing. Just turning it on and observing the performance is a great place to start. It tells me if I am playing in front of, behind or on the beat, just observing the transient markers.

The next thing I usually do, is then nudge the entire performance closer as a whole, by a nudge or two. Then I find areas that are still a little to loose for me and try to lasso groups of transients at once and manually move them to be closer. I always audition the movement right after making the edit to make sure I don't hear any unwanted artifacts to the edit. They occasionally happen for me, but usually if I try to move to much around, to far, all in one spot.

I have found that it does a pretty fair job at locking my rhythm section together. I have my Bass track located directly under the drum track and then rhythm guitar tracks directly underneath the Bass. This way I can visually see how all the rhythm tracks are lining up for timing.

Quantinize has never worked well for me to do all at ounce. The smallest note increment is 1/32 and auto Quntinize always seems to send the transient to the wrong 1/32, for me.

I've used it to loosely quantize multi tracked drums that were a basket case in some areas and it saved the day. It works well for rogue bass notes etc. The main use for me is for creating tempo maps for imported audio. If I have a click track I use that if not I use the kick track. I usually track to an Alesis HD24 and import the tracks into Sonar.
 
Rocky

The following is just observation / speculation / guessing on my part:
 
Whenever I set-up to use Audio Snap I resave my project (working mix) with a different name and then depending on what I am doing I either bounce to track and copy and paste that rendered track back into my working mix or export and then import the track back into my working mix.  Audio Snap, properly used is a very powerful tool but it gives me anxiety.
 
I wouldn't bounce to clip, I would bounce to track so I could listen to the render and then make changes and re-render if necessary.
 
If by bouncing to clip it is truly rendering and you are still getting artifacts, then my guess is that something else is going on ... it is not the bounce itself but how things are moved around.
 
Most of my experience is on drum tracks so here is an example from my experience that is probably not relevant but I will throw it out there.  Say for example I want to quantize a track and inadvertently ascribe transient markers to a series of 16th notes but set the resolution to 8th notes.  When I render, Audio Snap will force the 16th notes to line up where the 8th notes should happen and squish the notes together in an unmusical way.
 
The point is, the surrounding notes will influence the quality of your result and can result in weird sounding effects.
 
Other things that can cause weird effects are stretching a note too far or multiple renderings (that is why I bounce to track ... each evaluation is an independent rendering and when I am done the resulting track has only been rendered one time with Audio Snap).