I have been going through the Groove 3 videos on Sonar, and last night's "lesson" was on buses and routing.  The videos are good training on where to "point and click,"  but if one has no education/experience with recording (like me) it is fairly superficial knowledge.  
 
I believe I have a good idea on the use of buses (eg. - send groups of instruments to a bus for effects processing), but I'm a bit unsure of the reason for Sends.  As I understand, they can also have effects.  Can someone enlighten me on the practical uses of Sends?  I know that Anderton has touched on them before, but I can't seem to find what I was looking for.  
 
Thank you. 

I use buses to represent groups of instruments, e.g. the string section, the brass section etc.
This gives me one place to control section level effects such as reverb, compression, effects and section volume.
 
You could just as well keep all the individual tracks and duplicate the effects; this is however more CPU and labor intensive, and there is little lost is combining into buses.
 
Some effects, such as pan and possibly EQ, may be track specific and remain on the individual tracks. Reverb and compression generally should be on buses (imo).

Routing the Output to a bus (or aux) sends the entire signal on that path.
If you route a Send to a bus (or aux), you can specify the amount of signal.
 
Take a standard drum kit for example:
Most of us direct the Outputs from the Kick, Snare, HiHat, Toms, and Overheads to a Drum Bus for overall control.  We can change the volume as a whole, and add things like EQ and Compression to the entire kit at once.
We will then SEND each instrument (audio track) to another bus, which has Reverb (100% wet) in the FX bin.  We can then send a specific amount of each instrument to that Reverb Bus, so that it creates the ambient sound that we want.  You might not want 100% of the Kick Drum going to the Reverb, but may want 20% instead, while the Snare Drum could be sending 75% of the signal to the Reverb.  We then route that Reverb Bus Output to the Drum Bus, setting the volume sliders as desired to mix the kit with the room.

Side chain. :)

Can I chime in?
The problem I have with busses is that it's the same effect for a group of tracks. It sort of makes sense to me for drums. But still, its the SAME EFFECT spread across the entire kit, right?
Like the original post, I too don't yet understand how to effectively use sends and busses.

A send is exactly the same as the main destination selector. It just goes by a different name because it's considered an extra, or auxiliary path. You could call both of them "sends" and be correct. The only difference is you are not allowed to delete the first one.
 
The most common use of sends is for reverb. You want multiple instruments to share a common reverb for homogeneity and CPU efficiency, but each track will likely send a different amount of signal to the reverb bus. 
 
Next-most frequent application is parallel compression. You might, for example, send more of a lead vocal's double to the compressor than the primary lead vocal track.
 
Parallel distortion is another common scenario. I'll use send levels to send some kick drum to a distortion plugin, send a higher level to it from snare and toms, and usually none from the overheads.
 
Another frequent use of sends is setting up headphone mixes. You can have any number of headphone busses, allowing for a custom cue mix to every performer. Sends and send levels are how you determine what each person hears in his own headphones.

I use sends and busses all the time, in a creative way. Here is an example of a track where I have nine audio tracks and seven busses, (not counting the master buss):-
 

 
Probably not much use for "normal" music...

As bitflipper pointed out, in some ways the difference between sends and busses are semantics. What sends enable vs busses are:
 
1) The ability to use effect consistently, yet with varying amounts of wetness across different tracks. You could route all tracks to one buss and throw effect on it, but odds are slim you want the exact same amount of effect on every sound. The send allows you to have the consistency of exact same effects setting with varying amounts without having to duplicate instances of same effect.
 
2) In side-chaining, you can do a few cool tricks that would be more awkward if attempted via buss.  For example, let's say i have a Bass, a tuned 808 and a kick drum all fighting for space. Logically, the kick, having the shortest duration makes sense to use as source for sidechain source.One of the pitfalls of using the kick as a source is that any effects on kick may effect how sidechain is triggered .  Using a send instead of simple bus routing, you can send full kick signal, pre-send to a silent buss to use as source and still process kick in channel to taste with no consequences on side-chain trigger.
 
3) One of the biggest timesavers for me is setting up templates for particular genres and productions styles.  For EDM, hip-hop, and other predominantly electronic styles I use the following basic configuration to start:
 
Tracks
bass
Kick
snare
closed hh
open hh
perc
full drumkit (rarely used except when i compose entire drum part in third party program that has own sequencing like Geist)
synth
sytnh
rompler
piano
I add tracks as needed but this usually is a good start
 
Sends
Hall verb 
Plate verb
Ping-pong delay
Echo type delay
Exciter bus (pretty much a small chain that includes saturation and EQ presence bump to help a track cut through mix)
 
Busses
Drum summing bus (default bus for all drum sounds )
Instrument summing bus (default bus for all instruments)
Filtered bus (Drum,instruments and sends  are combined here)
Unfiltered Bus (this is used for any track I want to bypass all effects.  Don't use much, but I create it because the type of process i use this for is awkward to do after the fact.  I am struggling to fix a song now where I needed this and thought of it afterward).
Mixbus (Filtered and unfiltered combined)
 
By having my sends already configured, I save tons of time and i have verbs and delays already in place for any track I add after the fact.
 
 

Slugbaby and Bitflipper, clear explanations with real world examples. Thanks! I've been recording for a long time and you both just set things a good bit clearer in my mind.

And Dubdisciple too. Very informative post!