Here's how I handle my projects-
First of all I don't mix in Sonar so the only effects I use are guitar related( Amplitube, GTR, PodFarm, etc.).
 
When I have finished the recording phase I save as is. Next I bounce down all softsynths, then freeze the synths, and finally archive them. Now I save again as "Project archive". The last step I do is to delete the archived synths and their associated tracks and save again as "Project audio only". Now I'm ready for any eventuality. I burn my projects to DVD and delete from the computer...it's a lot easier to pop a DVD in than to try and read from a failed HD ;)
 
Bill

ADDENDUM TO MY EARLIER POST
A couple more reasons to print:

4. You may wish to open the project in another rig in which the relevant plugins are not present.
 
5. Several years from now, after a few successive upgrades, some of the plugins that you currently rely on will no longer work in their current form. I have learned this the hard way.

These days I usually only convert softsynth to audio when I need to do some kind of manipulation on the track that can only be done in audio such as reverse masking or slicing.

Side question. When I do convert synth to audio, I usually just freeze the track and copy the resulting wav to a fresh audio track. That method always seemed to work perfectly for me.  But from what I'm reading here, some folks actually bounce down to audio by recording during playback in real time. Is there an advantage to doing it that way?
 
Seems to me that by recording in real time you're putting your audio through a d/a conversion and then an a/d conversion. Which could open you up to timing issues as well as sonic issues depending on the quality of the converters you're using. I could see doing that if you were using some kind of high end outboard processing on the track. But otherwise I'm not seeing any advantage to flying the audio out of the box and then right back into the box. What am I missing?

JohanSebatianGremlin
These days I usually only convert softsynth to audio when I need to do some kind of manipulation on the track that can only be done in audio such as reverse masking or slicing.

Side question. When I do convert synth to audio, I usually just freeze the track and copy the resulting wav to a fresh audio track. That method always seemed to work perfectly for me.  But from what I'm reading here, some folks actually bounce down to audio by recording during playback in real time. Is there an advantage to doing it that way?
 
Seems to me that by recording in real time you're putting your audio through a d/a conversion and then an a/d conversion. Which could open you up to timing issues as well as sonic issues depending on the quality of the converters you're using. I could see doing that if you were using some kind of high end outboard processing on the track. But otherwise I'm not seeing any advantage to flying the audio out of the box and then right back into the box. What am I missing?

I don't think there are any D/A conversion issues by recording softsynths in real time, not with Sonar's recent synth recording functionality (perhaps you missed that announcement? You can now record a soft synth directly in Sonar). It's no different to freezing the track except that it's happening in real time so that you can do things like manipulate the synth's controls on the fly. The audio isn't going anywhere near your converters during the recording, it's all in the box. Unless you're talking about external hardware synths, but even in that case, the conversion is going to have to happen sooner or later, no? 

Hmm.... So is there a trick to setting up the routing to capture the synth audio before it hits the converters?
 
The only way I'd know to do it would be to route the output of the synth to my audio I/O and then patch that output to a set of aux inputs and route to a fresh audio track. Is there some other way to do it that doesn't involve the I/O box?

Danny
 
I freeze solely to keep track of what I thought was done. The next morning I thaw out those that need fixin.

JohanSebatianGremlin
These days I usually only convert softsynth to audio when I need to do some kind of manipulation on the track that can only be done in audio such as reverse masking or slicing.

Side question. When I do convert synth to audio, I usually just freeze the track and copy the resulting wav to a fresh audio track. That method always seemed to work perfectly for me.  But from what I'm reading here, some folks actually bounce down to audio by recording during playback in real time. Is there an advantage to doing it that way?
 
Seems to me that by recording in real time you're putting your audio through a d/a conversion and then an a/d conversion. Which could open you up to timing issues as well as sonic issues depending on the quality of the converters you're using. I could see doing that if you were using some kind of high end outboard processing on the track. But otherwise I'm not seeing any advantage to flying the audio out of the box and then right back into the box. What am I missing?

There is definitely no DA / AD conversion involved for VSTi. It's as simple has arming synth output tracks for record and hitting the record button. Everything else is handled by Sonar i.e. the signal never goes outside Sonar.
 
It is a major time saver compared to the sequentially executed freezing synths, which e.g. for multi-out drum VSTi can take much longer than just one song playback ... plus you can print your external synths at the same time ... while listening closely on headphone check for glitches, crackles, ...
 
Another possible advantage: unfreeze removes your WAV file (unless you manually copied it somewhere else) i.e. if after changing your synth settings, you don't have the previous one to compare to / comp from ... while with synth recording you keep the previous "takes" already properly grouped (if you have that setting enabled) ...

JohanSebatianGremlin
Hmm.... So is there a trick to setting up the routing to capture the synth audio before it hits the converters?
 
The only way I'd know to do it would be to route the output of the synth to my audio I/O and then patch that output to a set of aux inputs and route to a fresh audio track. Is there some other way to do it that doesn't involve the I/O box?

With a soft synth you can literally just set the input of an audio track to the synth's output, arm the track and hit record. Synth recording was added to Sonar about a year ago in response to years of demands. It doesn't go anywhere near your converters.

Rob[atSound-Rehab]
It is a major time saver compared to the sequentially executed freezing synths, which e.g. for multi-out drum VSTi can take much longer than just one song playback ... plus you can print your external synths at the same time ... while listening closely on headphone check for glitches, crackles

The thing is, if you have your buffer set low enough (which I believe you should do when synth recording otherwise the resulting track is out by a few samples - I've seen this with 100% quantized kicks from a drum synth) then you might hear crackles, but these aren't printed. When recording soft synths I set my buffer to 96, turn all other synths off and bypass all FX. Sometimes I will hear the odd pop and crackle whilst recording, but I don't hear these when I listen back to the track.

sharke
 Synth recording was added to Sonar about a year ago in response to years of demands. 

Ok so I'm not crazy. Until a few months ago I was on version 8.something-or-other. Seems to me at that time, the only way to capture VST synth audio was to involve the I/O box, which is why using the freeze function is my goto.