Physically speaking, sends and buses are exactly the same thing. Sonar calls them buses, so in reality a send is just a bus that's being used for an effect send. Pro Tools calls them "aux tracks," which I guess avoids the confusion. So a bus is a bus, and a send is also a bus. 
 
In terms of usage, a "bus" bus is used to group tracks together. So you might route all of your drums to a drum bus or all of your guitars to a guitar bus. This way, not only can you control the volume of all of these tracks with one fader, but you can also apply effects to the whole group. For instance, you might use some gentle bus compression on a drum bus to "glue" the parts together. 
 
A send is just a bus that's being used as an FX return. So for instance, you might set up 3 different reverb sends to cover your whole project, e.g short ambiance, hall and plate. Or you might set up a couple different delay sends, e.g. slap delay and dotted eighths. 
 
It's worth remembering that you can send a bus to an FX return as well. Let's say you have a vocal bus which you're using to control the overall level of 5 vocal parts with one fader. If you require the same amount of reverb on all of these vocals, you can set up a send from the vocal bus to the reverb bus.
 
You don't have to leave the send level at 100%. You can set it how you like. Since you've (hopefully) set the wet/dry control on the reverb bus to 100%, setting the send level to 100% is going to give you a LOT of reverb (especially if you have a long reverb like church or cathedral) which may or may not be what you want. 

sharke
Physically speaking, sends and buses are exactly the same thing. Sonar calls them buses, so in reality a send is just a bus...etc etc

lol--JKO's confusion called for some simple clarifications, so the screenshot will be helpful.  We're using Sonar - Sonar calls the knobs Sends, and the channel strips on the right Buses, so - it's helpful if we use the terms our program uses. 
 
Randy

Do you want the sound from the perspective of someone in the back of the cathedral? Then directing all the tracks' main outputs to an effects bus is fine. If you want to make your sound more intimate, or close up, the individual performer groups would be treated with their own reverb mixes.
 
Sometimes you need to change all levels at once, sometimes individually. One way to control that is by putting controls into groups so they are linked together. Then moving one of the controls moves them all together, but by holding the CTRL key, you can change just one of the controls without changing the others.

Aside from terminology, sends and buses are not the same thing.
 
A bus is a stereo (or surround) channel to which audio can be routed. A send is one of the ways of routing something to a bus. The other way is via a track or bus output.
 
Not to be a pedant, but I think this is a distinction worth being clear about. A bus is an actual thing. Send and Output assignments are more like instructions.

Sure, this makes sense, I can route all my vocals to a vocal bus and then route that bus through a send. Btw, for any IT geeks out there, bus management sounds a lot like Active Directory group management, doesn't it.
 
So right now, my vocal tracks are all routed through Vocal Send which has my reverb on it. The output is also Vocal Bus for each one. I should just route the Vocal Bus through the Vocal Send. Does that provide any CPU resource savings as well, just curious?
 

jkoseattle...I can route all my vocals to a vocal bus and then route that bus through a send...I should just route the Vocal Bus through the Vocal Send. Does that provide any CPU resource savings?...

It saves power to use one reverb plugin for all those tracks, instead of a separate reverb instance for each one.
 
I think you have it straight now, but I pulled out your quote because you said "...route the vocal bus through the vocal send..." - Actually, as per the screenshot posted earlier, you're routing the vocal bus to the Master bus.  You're routing just a percentage of its signal through the reverb bus via the Send knob on that vocal bus.  You probably have it set up right now, your description was just inaccurate so wanted to be sure.
 
Randy

CPU saving is due mostly from having less FX not from how things are routed. Using a send is used to apply FX to possibly more signals then one FX bin in one track thus saving on the FX. But as stated a send is not the only way of doing this. You could output a track or tracks to a buss just as well. It all depends.
 
How one uses sends or whether they are used at all depends on what is wanted from doing so. Think of a send as a tap on the signal stream of a track that can be sent to a buss for its own processing. It can be pre fader or post fader. Each has an effect on how the signal will be processed.
 
So far we have been talking about a single send going to a single buss.  But a track can have multiple sends going to multiple auxiliary buses.
 
Though it seems simple it really isn't and its up to the user to fully understand what  sends can do.  

Its only wrong, if it doeskin sound the way you want it to sound. Its only good, when it sounds the way you want it to sound. 
How you get this is up to you as there are over 1,000,000,000,000,000,000,000,000 Quintilian way to get from A to B. all you need to do is learn how to use the effects and tools you have and then you will be able to use your ears and knowledge to dial in the sounds you want.
 
As far as your volume faders and sliders. this is called gain staging and there are many ways to gain stage when recording and mixing.
Basically things go like this:
1. Tracks go to the Master bus
2. Sends go to buses, if you use sends. Pre send is before the track fader and Post send is after the track fader.
3. Buses go the the master bus
4. Master bus goes to the main outs 1/2 of the sound card.
 
Now your master bus, buses and your main outs should stay at 0dB (Unity Gain). You control the volume when mixing with the track faders and the send levels (if you use sends), everythnig else is set at 0dB (unity)
 
See that is pretty simple as it works just like a hardware mixer. People think since its on a PC, it will be different , but its not!!!
 
CJ

Setting the master bus and outputs to 0Db makes some sense at an intuitive level I guess, but I'd like to know in writing why this is. Is it because it's the way to ensure the cleanest STN ratio? Is it analogous to a trumpet, where the skinny part of the pipe is where the nuances of sound are crafted, and when the sound reaches the end of the tube, all manipulation has taken place, and so there's a big wide bell that basically says "Whatever was done to the sound as it made it's way through the pipe, not take that sound and make it as loud as it can be"?

You are over thinking this. Its just convenient. It insures that you have a standard to mix to. If you vary the level on the master buss then you wont know what your tracks are really doing. The level on the master buss should be the sum of all your tracks and buses.