Since a few months, I watch a mysterious behavior after splitting audioclips.
I mark the whole clip and split it at a certain point.
In preferences I set: Selection after single split: "left portion".
So the left portion should be complete deleted - but every twentieth time a tiny bit of the audioclip remains at the left side. This tiny bit is so short (0,01 seconds), that is almost unvisible. 
I have been working for about 20 years with Sonar, earn my daily money with it,but I have only experienced that behavior since a few months.
This has effect to the clip when editing the visible main piece. Because Sonar holds the whole audiodata, even after "apply trimming" to the main piece, and references to the unvisible tiny bit. I also watch sometimes corrupt audiodata when normalising or applying an effect to an audioclip. I'm not surprised if problems occur by this behavior.

 
I also posted this for a few weeks. Mybe that is relateted:

These are very basic things that should work.
Every day I experience it and have to take attentions to this misconduct.

Regarding the first one, I've had something similar bite me when I snapped to zero-crossings. If the zero crossing was not on the beat and then I had another take that was snapped to the beat, there would be these tiny overlaps when cutting if the Take Lanes were folded up and I was working on the parent track. That may not be the cause of what's happening with you, but I end up with the same results.
 
Regarding the second, the people at Cakewalk are aware of this but can't reproduce it. I'm waiting until it happens in a project so  I can send the project to them, but it hasn't happened in months.

It is unfortunate to see a daily bread and butter user get a canned response of "no endpoint in sight." When companies who take pride in product rely on vendors who don't, decisions get made accordingly.

Because we are digitazing analog recordings we have many terabytes of audiodata. Therefore we observe such problems more often. Some projects have up to 80 GB audiodata. I don't want to do this in Wavelab or Adobe Edition.
Sonar is unbeatable and the best and fastest software for such audio projects.
But sometimes it is sad when the new updates also make some basic things flawed.
 
 
 

Just for testing purposes, next time this happens, hit undo to start where you were previous.
Lasso the whole clip (to make sure every part of the clip is selected.
Selected "Bounce to clips" This will ensure there wont be any stray pieces of your audio clip hiding underneath the clips.
 
Now proceed to do the split again, check to see if this worked.
I have had similar things happen to me where slivers of a clip were hiding underneath a larger portion of a clip.
 
Please remember im just trying to help and not deny the fact you might have a different issue.
This might be something totally different, but I did take the time to write out these steps, so please just take it for what it is. Someone just trying to help.

mettelus
It is unfortunate to see a daily bread and butter user get a canned response of "no endpoint in sight."

Are you referring to my response? If so, I'm simply another bread and butter user describing my experience. I'm a) confirming both phenomena, b) mentioning how the first one has happened to me and what caused it, and c) confirming that I've had the second occur but cannot reproduce, nor can Cakewalk. Cakewalk can't fix something they can't reproduce. Equating "no endpoint in sight" with "people are trying to reproduce and solve the problem" makes no sense. It's not my fault when something happens so rarely I can't summon it on command.
 
P.S. I do think it relates to mono files because I think it happened when I was working on narration. But this was almost a year ago so I can't be 100% I'm remembering the circumstances correctly. So you might think it's pointless for me to say anything, but if THambrecht reads this and the next time the second thing happens he was working on mono files, then we're starting to get somewhere...Cakewalk can then concentrate on trying to reproduce with mono files if that's a more common error mode (yes I know his screen shot shows stereo, but that may or may not be relevant to the cause...I don't know what the history of that waveform was).

In our case it happens daily.
I also cannot reproduce it. I don't even notice when such a bit of audioclip remains. It's nearly unvisible.
When I mark all clips and export hundreds of clips, I see a few exported wav-files with only 5 kB of length.
I thought it was related to the fact that we have clips with 1 or 2 hours.
 
But I try to figure it out. I'm trying to reproduce it.
 
 
 

THambrecht
In our case it happens daily.
I also cannot reproduce it. I don't even notice when such a bit of audioclip remains. It's nearly unvisible.
When I mark all clips and export hundreds of clips, I see a few exported wav-files with only 5 kB of length.
I thought it was related to the fact that we have clips with 1 or 2 hours.
 
But I try to figure it out. I'm trying to reproduce it.

 
Is the sliver in a different Take Lane?
 

and you don't think "Bounce to clips" wont take care of this ?
 

chuckebaby
and you don't think "Bounce to clips" wont take care of this ?

 
I think it probably would, but I'm still curious how the "microclips" are being generated in the first place in case they're coming from someplace unexpected. For example, I wonder if Split will snap to zero crossings if that's selected, thus leaving a sliver behind if the clip extends beyond a zero-crossing. Don't have much to go by, so grasping at straws.