Here's a question for the computer guys in this forum.  I'm a producer/guitarist and not a computer guy.  I've been recording lately at 96/32 in Sonar.  My hardware doing the analog to digital conversion is either a Mytek 96 A/D, RME MultiFaceII, or Roland VS700.  All of these support 96/24.  I'm recording 32 bit in Sonar for the sake of not being concerned about transients peaking at 0.  Anyone see any issues with this?  It's sounding good but I just thought I'd ask.

24 bit is appropriate. 32 only on internal bounces.

Typically, all the better any converter can do is 24 bits...so you can't record at a resolution better than that.  Internally, SONAR will handle everything as 64 bit floating point (if you have "64-bit Double Precision Engine" enabled), so headroom isn't an issue.  So...saving individual tracks as 32 bit doesn't really buy you anything.  It just costs you disk space.

the only thing you may want to think about is not so much the bit depth, but the sample rate. the nomenclature works like this BTW: 24(bit)/ 96(khz). If your soundcard is only capable of doing 24 bit, that's all you are getting. 32 bit values beyond the 24 are not filled (or I think more accurately, filled with 0 values). Now, on to the real deal. If you plan on distributing your stuff (even just burning disks on your computer) you have to adhere to the format that you are using 24/96 is DVD, 16/44.1 is CD. (there are a couple of other configurations, but they are less common)
The sampling rate , if not handled correctly (and I don't think Sonar is capable of converting it)
will yield in errors if you try to burn a CD at 24/ 96khz. You can put 24/96khz on a DVD, but you'll have to listen to it on a DVD player.
Now here's the kicker. You can record at any bit depth you want (24/32/16, whatever) and dither it to 16 bit for CD distribution. It's just the sampling rate that is the sticker here. I think you will find a lot, if not almost all folks around here use 24/44.1 just to avoid the sampling rate conversion.
I know this is an oversimplistic view of how "it works" but I hope this gets you pointed in the right direction.

Also remember that no 24-bit converter delivers a true 24 bits. Converter technology is just not that great.

so I gain nothing- really- by having Sonar set to 32 bit in the File Bit Depths preference?  Okay, but have I been LOSING anything, in anyone's opinion?  I just converted a song I'd recorded in Sonar at 32 bit to 24 and I may be imagining but it seems slightly better!  Am I tripping?  Ha!  I'm most appreciative of any responses!

You are losing disk space. You are also tripping.

What matters is the 64-bit audio engine for doing calculations. 32 bit file depth simply allows you to reproduce irrelevant data with more precision 

...think you're right that I'm tripping...  ha!  I appreciate the insight from everyone!

Anderton
What matters is the 64-bit audio engine for doing calculations. 32 bit file depth simply allows you to reproduce irrelevant data with more precision 

This is the very best way of putting it. I have to steal this when this subject comes up again.