Howdy.  Given the routing flexibility in Sonar (and most other DAWs, I'd imagine) there are a bunch of ways a person could set up a headphone mix for tracking vocals, just interested in hearing how other users set up their vocal sessions.  My method may be a little convoluted, but it seems to work pretty well:
 
I typically track vocals last, so I send all music tracks into instrument buses (drums, guitars, etc.) then create a send from each of those buses into my headphone output.  The armed vocal track is routed directly to the headphone output, and I also set up a send to a fairly hyped reverb (the output of that reverb bus is also sent to the headphones.)  That way I can get a nice balance of my live voice and backing tracks while recording.  For some reason I can't get my interface (Saffire Pro 40) to send the mic input directly to my headphones, so I need to enable input echo on the track within Sonar.  Works pretty well, although things get a little messy when I start tracking backing vocals, because if I forget to turn down the reverb send on previously recorded tracks, the mix gets pretty splashy in the headphones. :)
 
At any rate, that's how I do it.  Any other tried-and-true approaches y'all use?

I use a send from the MASTER Bus for the headphones.
The UFX (RME TotalMix) lets me route the Vocal Mic straight to the phones.
No latency issues, etc. (gotta love RME and TotalMix).
TotalMix has Reverb and Delay I can add to the incoming Vocal (sent to phones, not Sonar).
Bummer the Saffire won't let you send the Mic to Phones...
Tom

DeeringAmps
I use a send from the MASTER Bus for the headphones.
The UFX (RME TotalMix) lets me route the Vocal Mic straight to the phones.
No latency issues, etc. (gotta love RME and TotalMix).
TotalMix has Reverb and Delay I can add to the incoming Vocal (sent to phones, not Sonar).
Bummer the Saffire won't let you send the Mic to Phones...
Tom

Pretty sweet that you do can do all input monitoring and effects out of the box!  The Saffire does allow for direct input monitoring, but I've never been able to get it to work.  It doesn't have any effects or VST support though, so I'd have to go through Sonar for the reverb anyway.  I set the ASIO buffer to 64 samples whenever I'm tracking and I don't notice the latency, so at least from that perspective my current methodology works fine. 

streckfus
Howdy.  Given the routing flexibility in Sonar (and most other DAWs, I'd imagine) there are a bunch of ways a person could set up a headphone mix for tracking vocals, just interested in hearing how other users set up their vocal sessions.  My method may be a little convoluted, but it seems to work pretty well:
 
I typically track vocals last, so I send all music tracks into instrument buses (drums, guitars, etc.) then create a send from each of those buses into my headphone output.  The armed vocal track is routed directly to the headphone output, and I also set up a send to a fairly hyped reverb (the output of that reverb bus is also sent to the headphones.)  That way I can get a nice balance of my live voice and backing tracks while recording.  For some reason I can't get my interface (Saffire Pro 40) to send the mic input directly to my headphones, so I need to enable input echo on the track within Sonar.  Works pretty well, although things get a little messy when I start tracking backing vocals, because if I forget to turn down the reverb send on previously recorded tracks, the mix gets pretty splashy in the headphones. :)
 
At any rate, that's how I do it.  Any other tried-and-true approaches y'all use?

This is EXACTLY what I do, minus the reverb.
Each bus feeds into a Pre-Master buss which normally sits at 0dB. Then I simply lower the fader until I get the correct balance between voice & instruments in the headphones prior to recording.