I'm trying to determine which is the best way to route tracks to busses. I can use a send from a track or I can simply set the track output to a buss. I typically just output the track to a buss and can't quite figure out why I'd want to use a send.
 
I'd love a bit of education on the benefits.

There are many ways to approach this. I am pretty "old school" so this is how I always do it.
 
If I create a bus as a sub mix, I route all corresponding tracks to that bus. EG: An electric guitar bus will have all electric guitars routed to it. A BG Vocal Bus will have all background vocals routed to it, etc... etc...
 
If I create a bus as an effect (Long Reverb, Short Reverb, Long Delay, Short Delay) then I will use "sends" to access each bus. The track itself will be routed to it's appropriate "bus track" (as mentioned above) but I can send signal to any of the "effect bus tracks" using sends. This gives every single track in the project access to any effect bus.
 
Sends also work well for parallel processing... but that an entirely different topic.

Perfect. That's exactly what I was hoping to hear and it makes perfect sense. Thanks.

Yea, what Sean said.
 
Along the same lines, I'm wondering what folks do with the outputs of their effect return busses? I have problems sometimes in that usually I send the outs of these busses to the master, But, if there is any processing on the other instrument subgroups, I tend to get artifacts, or oddness, out of the effect busses because the sends of those instruments usually come from the track sends . Hope that makes sense.

codamedia
There are many ways to approach this. I am pretty "old school" so this is how I always do it.
 
If I create a bus as a sub mix, I route all corresponding tracks to that bus. EG: An electric guitar bus will have all electric guitars routed to it. A BG Vocal Bus will have all background vocals routed to it, etc... etc...
 
If I create a bus as an effect (Long Reverb, Short Reverb, Long Delay, Short Delay) then I will use "sends" to access each bus. The track itself will be routed to it's appropriate "bus track" (as mentioned above) but I can send signal to any of the "effect bus tracks" using sends. This gives every single track in the project access to any effect bus.
 
Sends also work well for parallel processing... but that an entirely different topic.

Exactly what I do too.

back in the old days when all I had was a 4 track recorder, this was really the only option.
you had 2 sends on each track. All outputs of those 4 tracks went to the master bus.

as you can see in the example above. going from bottom to top= there is a volume / pan / 2 sends
on each track. don't be misunderstood this is NOT an 8 track (5-6+7-8) Those last 2 faders (to the right)
were stereo inputs used in conjunction with the 2  1/4 inch jack sends located on the back.
in other words, you hooked up your hardware's outputs to 5 thru 8.
-The sends controls the amount of signal that goes to your hardware (the input).
-The faders controls the amount that comes from the hardware (output).
 
so now we go to a realistic approach in something like Sonar.
you typically never use a send for something like compression. unless of course your using it as an effect.
IE= Parallel Compression. 
You will however find a send is very productive in FX like Delay, Reverb, Chorus, Flange.
 
Example=
lets say I have 3 vocal tracks (1 lead vox / 2 background vox) 
I want the lead vocal to have a good amount of reverb but I don't want the background vocals to have as much.
(you want to be able to distinguish the tracks. putting the same amount of reverb to all 3 vocal tracks will make them muddy and dare I say: blurry.)
there is no way possible to accomplish this using a bus. your only other option is to put a single reverb FX on each individual track. which lets face it, that is chewing up CPU.
so what we do is:
 
1-create a bus.(set the bus output to the master)
2-insert a reverb in to the FX bin of the bus.
3-insert a send to each of the 3 vocal tracks (the 1 lead vox / 2 background vox).
4-turn the lead vox send to around 5 o'clock.
5-turn the 2 background vox sends to 3 o'clock.
 
taking it one step further, you can automate the sends. which IMO is one of the best parts
about using sends vs. a bus.
if there's a long vocal note. apply automation to crank up the send during just that one part.
of course all the while being thoughtful and aware of what too much reverb does (drowns the sound).
 
Sends are critical in saving resources and mastering the tool.
 

Sends also give you additional controls of the levels, in addition to pre/post fader. Not sure if that was specifically hit above, but dovetails to side-chaning (ducking, gated reverbs, yada yada).