I crop all my tracks. One by one. Edit out all noises and apply crossfades front and back. It is my understanding this is common practice in the industry. I have been reading books where renowned mixing engineers explain this.
It is non destructive because like Karyn stated you can grab the end of the clip and drag it to the right and or left and everything you cut out is still there.
Using crossfades correctly can help clean up/smooth out sibilance and plosives as well. Crossfades are recommended when cutting clips like this because to eliminates any clicks, pops or other noise that can happen during play back when it goes from silence to the clip and the clip back to silence. Crossfades are applied at the beginning and end of every clip no matter how big or small. Its what the pros do.
Andy's (Synkrotron) screenshot is a great example and demonstrates this very well.
 
Cheers.

god-z
It is non destructive because like Karyn stated you can grab the end of the clip and drag it to the right and or left and everything you cut out is still there.
 

 
not if you do what i do.
 
'apply trimming'.
 
then it is gone forever.
 
like i said.......... "EXTREME PREJUDICE"
 

Because there is now a rumor that Cakewalk might be adding a wave editor at some point then it would add the option I have used in the past.
I tool copy the track to Wave lab and then it's pretty easy to silence the silence.
Drag to highlight,  hit backspace.  This is super fast and I often have used it on tracks that had unwanted noise between the good parts.
Of course it's destructive, I don't care,, good riddance to bad rubbish I say. 
There's always a back up anyhow. 
It will be interesting to see how Cakewalks future wave editor compares in ease of use to Wave Lab. 
 

I tend to use track automation envelopes, and I rarely take them right down to -INF just to retain whatever"air" which might be noticed by it's absence

I also tool copy and destructively edit often (Adobe Audition 4). In my perspective, if it is truly noise I want it gone for good. This takes less than 30 seconds per wav file usually.

I also freely admit that surgical modification (aka automation) is way beyond my patience threshold.
 
Edit: I forgot I posted this last year which is an example of how I typically NR pass audio with Audition: https://youtu.be/aclxPcBEQEc  That particular file is interesting since it had noise that ramped up during the take (had a nice ground issue going), so the noise capture was actually done on the end tail when used in my project, but in this video I grabbed the middle so you can see the residual noise on that end tail. The free version of Audition 3 has this same capability.

Having read the many new comment here, I think I will stick with my current method of splitting and trimming audio clips in track view, applying fades and cross fades where needed. I find this system quicker and easier than automation, and instantly visible without having to open an automation lane or changing track view to clip automation...

Cactus Music
Because there is now a rumor that Cakewalk might be adding a wave editor at some point
 

Source?

That question was specifically on the survey that went out a week or so ago, IIRC. Several features from the Feature Request forum were in that survey. I swore I saw a page referencing up and coming such things, but cannot find it and forgot who/where it was linked now.

I will use one of several methods.  Either work fine and I use them both.
 
There are times I will go through a track....especially the vocal track lead....I move it to the top of the track view and where I have noise from lip smacks or any other noise, I simply drag on the time line above the track after I have highlighted and selected that specific vocal track.  The section of the track that was dragged on turns a darker color. Now I use Audio>Process>gain>mute to totally mute the noise. By zooming in, you can get very precise with this "trimming" method.  If you goof up, you can UNDO and back up. Once you save and exit however, it does become a destructive, can't go back,  edit.
 
Note: You can also use this same method to increase the volume of a specific word or part of a word..... say for example the singer sang a word correctly but for some reason, backed off the mic a bit and the first or last part of a word is almost inaudible.  Simple zoom in and use the cursor to highlight just that one part of the word.... use the Process>gain>+3db ( or some other amount) to bump that part of the word up. You might have to play with it and bump it several times.... but you can easily rescue the word rather than having to punch to fix it.  I use this little trick quite a bit.
 
Another way I work is that I always use envelopes and add my nodes manually and surgically. I'm going to use envelopes and since they are there, I can use them in place of the first method. Some edits are instant and others are slight fades in or out... but quick.  See the picture below. By zooming, the nodes can easily be placed in a surgical manner, exactly where you want them.
 
BTW: if you are using either of these methods, you will not get POPS in the edits.  These are silent edits.