I have some 8.5 sessions that were recorded in 44.1/16 bit that I have to remix/remaster.  Is it best in the way of audio fidelity to just leave them in the format the tracks are all in now, or try to upsample them to 48 24bit (which is what I use now for all my more recent stuff)?  Thanks for any advice.
 
When I originally did these recordings, I wasn't aware of the sonic hash that might result mixing down 44.1/16 tracks down to a master track at 44.1/16.  Also, my older system wouldn't handle 48/24. 

that's a 44/16 project, leave it that way, imho. i can see no advantage to upsampling tracks to 48, unless you need the individuals tracks for vegas or premiere, say.  
 
when you export, export the mix to 48/24 if you need to. 
 
i export 44/16, then upsample or mp3 in sound forge.

I'd leave them at 44.1, as there is no particular advantage (or harm - Sonar's sample rate conversion is very good) in upsampling them. 
 
And in terms of bit depth, Sonar does all its processing at a higher internal bit depth anyway. The various bit depth settings only affect what happens when files are written or the signal is output through the sound card.

From everything I've read I'd say just leave them as is. Not sure what extra benefit, if any, there would to be had by doing this but considering you'd be rendering back down anyway which removes stuff you might actually lose stuff that's already there. Also I've seen lots of discussions that the difference between 44.1/16 vs higher sample/bit settings isn't that perceptible by most listeners anyway.
 
However... if you were doing some overdubs on the tracks I could see the benefit of bumping up the bit/depth of the original files to record the new tracks to and apply effects/mix/etc at those settings then mixing the new tracks down to match the original settings then and using them in the old project (and scrapping the tracks you upsampled).
 
That is purely speculative though and I am by no means an expert. Cheers.

Beepster
However... if you were doing some overdubs on the tracks I could see the benefit of bumping up the bit/depth of the original files

You bump up the bit depth of any file that isn't already in 32 bit floating point (or higher) just by loading it into Sonar. Sonar processes everything internally at either 32 bit floating point or 64 bit floating point (if the double precision option is selected), regardless of the file's bit depth.

drewfx1

Beepster
However... if you were doing some overdubs on the tracks I could see the benefit of bumping up the bit/depth of the original files

You bump up the bit depth of any file that isn't already in 32 bit floating point (or higher) just by loading it into Sonar. Sonar processes everything internally at either 32 bit floating point or 64 bit floating point (if the double precision option is selected), regardless of the file's bit depth.

Interesting. Excuse my ignorance but if that is the case does it simply not matter what my settings are while recording? I do notice you only mentioned bit depth but not sample rate. Seriously, not disputing this (still very much a n00b)... I would like to learn more about this and how it affects a) recording, b) mixing and c) mastering or any other things I may be overlooking.
 
Cheers.

The bit depth setting for recording affects how many bits are used in the files written by Sonar when recording. There is no reason to use a bit depth here that is different than the audio interface is set to.
 
Sonar processes internally using 32bit floating point (or 64bit, if using double precision). 
 
Sonar also has other settings for bit depth for exiting Sonar - not just exporting, but for rendering (like when freezing or bouncing tracks).
 
Plugins may use 32bit floating point, 64bit floating point or something else, regardless of what Sonar's mix engine is using.
 
 
In terms of sample rate, it's more complicated. Sometimes it's advantageous to do certain types of processing using a higher sampling rate. Generally, plugins upsample internally as required when this is the case, but there may be certain plugins that don't (especially old ones). You also might get lower latency at higher sampling rates.
 
But in terms or recording/playback, higher sampling frequencies only offer higher frequencies. If you already can't hear any loss of high frequency information at 44.1kHz, then a higher sampling frequency offers you nothing. People who think higher rates are "more accurate" (aside from offering higher frequencies) simply don't understand the basics of digital sampling - even if they think they do.

Thank you for taking the time to reply. I am reading that over and over again parsing it out and attempting to digest it. I think I'm getting it. Gotta go do something else for a bit though but I'll be back. Again, thanks.

drewfx1
..... You also might get lower latency at higher sampling rates.

I think you meant to say higher latency.
 
....edit I was wrong about this you are correct.

When I move to higher sample rates my reported latency goes down. I though that was how it worked.