so this may seem like a noob question (prob is), but OK, I understand dithering is meant for when you are reducing bit rate.. say you are exporting a 24 bit project to a 16-bit file for CD burning, but why can't I just export as 24 bit and then convert it outside of Cakewalk using an audio conversion program?
 
I guess i just dont understand why dither?  Also should I be dithering as I work in the "Playback and Recording" area in preferences?
 
Thank You,
Adam

Here is one explaination.
 
When you move from 24-bit to 16-bit, the possible amount of values contained in audio recording is reduced. A bit like MP3 reduces the quality by filtering and removing frequencies.
 
Dithering adds a discrete Pink or White Noise to the audio track before reducing the bit rate. This compensates by using a noise shaping technique to psychoacoustically fill the void. Without this you would just remove anything that couldn't be kept within the bounds of 16-bit.
 
Since you won't always be using 16-bit, it's good 24-bit as a default.
 

Cool thank you.  What if I am exporting my music to a WAV file, then convert it to MP3 outside of Cakewalk?   Do I need dithering then?   Is it just for bouncing to 16-bit?
 
Also do I need it on in the prefs or just while exporting?   
 
Thank you interpolated!

One explanation that I read that makes sense to me:
 
Like said, the conversion to something with less quantization steps introduce artifacts - many values in 24 bit become one value 16 bit.
Ears/brains makes patterns of everything that occur in the same spot - like a certain quantization step.
Adding noise makes a little randomness to this very spot - sometimes it becomes a ONE, sometime a ZERO.
So less for brain to make patterns off.
 
In short it sound better to our ears.

To answer your first question, you can export a 24-bit file and convert it in another program. Just bear in mind that not all programs will yield the same results, nor have the options available that SONAR does for that conversion.

You only need to worry about dithering during export (if needed).

If you're working with 24 bit audio then turn dithering off until you do the final actual conversion to 16 bit.

Lots of other applications can convert 24 bit to 16 bit, however most of them will (or can) add dithering of some sort or another.

As for "why dither", some people can tell if audio hasn't been dithered some of the time, others seem not to be abke to tell dithered from undithered. It seems to be very dependant on the original audio and the algorithm used. Dithering is a very subtle thing and may not even be noticeable. If in doubt export several times, once without dithering then repeat suing the various dithering algorithms Sonar provides then see if you can hear any difference and go with the one that sounds best.

tlw
If you're working with 24 bit audio then turn dithering off until you do the final actual conversion to 16 bit.

This isn't necessary. SONAR won't apply dithering to any bounce/rendering process unless there's a reduction in bit-depth.

I export 24 bit files to Sound Forge without dithering, edit the stereo file in SF, and "save as" to 16/44.1, using a template, when I'm done.  SF does an excellent job of dithering.a  I could just as easily export to SF at 16/44.1 and use Sonar's dithering, but that's what I bought SF for.  Let your ears be your guide.

Most of the answers given that try to explain dither are wrong in whole or in part. They may be easy to understand but they are wrong.
 
The short answer:
 
Dithering is just adding random noise at roughly the level of the lower bit depth step size before reducing bit depth and it does two things:
 
1. It makes the often ugly quantization distortion sound like noise - essentially because the distortion becomes based on a random signal (the dither) instead of your signal.
 
2. It preserves part of the signal that's below the lowest level possible at the lower bit depth. Instead of the very quietest parts being cut off completely, they are still there to the extent that you can hear them in the noise.

truncation introduces artifacts that are statistically determinable. dither breaks up the statistical determinability of these artifacts. if you ever have the opportunity to view the spectrograph of a 24bit sine wave truncated to 16bit without dither and truncated to 16bit with dither you will immediately know what dither does and why you need to apply it.