Hi,
I know this a bit subjective, but I'll try and describe as best I can.
Listening to a recent mix, I find it a little harsh, not as in too much top end
just a bit harsh, I'm trying to smooth it.
For example, on the master bus, I've put sonitus EQ, and taken the middle (800 Hz) fader
and pulled it left, to about -15 dB, and then pulled the next two to the right, to about 8dB, to try and get some
top end, is there another way to do this. I hope this makes sense.
 
Thanks
John 

Some light compression?

I export my mixes to a 24 bit stereo wav then bring them into a new project. There I use the fancy arsed EQ and compression right on the track to get everything nudged a little closer to where I like (this is ever so subtle) and now that I have the PC2A I put that on the master and set it to apply a little more light compression. Seems to work alright for making things a little less hairy... and believe me, my tunes are hairy.

Oh... and that track EQ compression is the LP64 or whatever it's called. I use the Sonitus in the original project or just stick to the PC eq module. 

Keep the volumes lower in the tracks instead of relying on limiters in the buses. Make sure the EQ is not coloring the sound.   

If I were using that sort of EQ on your masterbuss I would be concerned. That is not good. What you should be doing is reviewing the tracks one by one and listen to all of them and try and determine which ones are causing the harshness. I would aim for a nearly flat EQ on the masterbuss (or no eq at all) and get the total sound happening at track level. Much better way to go. That way too you have got more control over individual elements rather than effecting everything as a whole. 

1) why not try a little tape saturation on the master bus OR just the tracks that sound harsh. That's if the Cakewalk one works for you without crashing, lol. There are others though like Massey Tapehead (now available as free VST although only in mono)
2) Are you using console emulation? I find this works well in reducing digital harshness, if that's what the problem is. 

Post #5 is right on.

Another thing I like to do is, use Subtractive EQ.

For example, if you want to increase the 1kHz area, use a High and Low Shelving and lower the frequencies on both sides of it. It takes some more work, but I always find I like my mixes more when I use this method.

This needs to be done on a track by track basis, and you'll find by the time you get to the Master bus, like Jeff said, you'll need little to no EQ'ing.

Do all this before your compressors and limiters as it will completely change how you set those up.

These some good video's on u-toob about Subtractive EQ'ing.

Another thing I like to do is, use Subtractive EQ.

For example, if you want to increase the 1kHz area, use a High and Low Shelving and lower the frequencies on both sides of it. It takes some more work, but I always find I like my mixes more when I use this method.

I read somewhere that this is because boosting frequencies creates artifacts, while cutting them does far less so. Although I could be wrong. 

sharke

Another thing I like to do is, use Subtractive EQ.

For example, if you want to increase the 1kHz area, use a High and Low Shelving and lower the frequencies on both sides of it. It takes some more work, but I always find I like my mixes more when I use this method.

I read somewhere that this is because boosting frequencies creates artifacts, while cutting them does far less so. Although I could be wrong.

Yes, it's something to do with aliasing or something like that, I forget the 'technical' reasons behind it. The other thing is, you are just simply increasing anything unwanted that you may have captured in the recording if you boost. It all seems to melt together so to speak when I take the extra time to do it this way. Probably all in my head though.

My approach is this:
1) Sub-buss each instrument (i.e. Guitars, Bass, Drums, Vocals, etc.
2) Solo each Sub-buss and just listen. find out which instrument seems to be making it worse. Or, conversely, Mute one sub-buss at a time and see which instrument seems to make the 'Best' impact by being taken out. 
3) Work on each one individually to see what frequencies are harsh. To do that, I use a Linear-Phase EQ, such as DMG EQuality (not the EQ you're already using in a track or buss--In fact, "Power" those off.) Using a single band, tight Q, and about 10db of boost, sweep the area you think is harsh and find the center frequency that seem to be the worst. Stop the sweep there, and then cut that frequency by about 3-6db (or more, if you need). Then, re-enable your normal EQs and unMute/unSolo your busses and see how this goes. 
4) Repeat. 

You may also add a little "air" up top with an EQ Shelf around 10Khz+ to help your ears.

Everybody else here has some great input. Don't just follow one method. Try them all or even combine methods. If all else fails, if one instrument is still bad, re-track it. I prefer to fix the source than fix the mix, where possible.