upsample previous session tracks from 44.1 to 48 good idea?

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. 

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.

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.

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.

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.

Yes, all things being the same the latency will drop with a higher sample rate because it takes less time to fill the buffers.

Veeery eeeenteresting.