Okay when thinking about mixing and EQ never lose sight of the purpose--which is to create an intelligible mix with clarity and power. Myself I have an approach that may be a little bit more radical but has served me fairly well.
First off I'm a big believer in using shelf filters to nip and tuck sounds. I use a LOT of high pass filtering to roll off bass frequencies on almost every instrument. For all practical purposes I filter everything in some way or another.
I usually run a high pass filter to eliminate anything below 100hz on guitar, snare, toms and so forth. For cymbals I usually start the cutoff around 500hz. Vocals about 150hz or so. The reason I do this is I only want the bass and kick drum occupying the space below 100hz to allow for a powerful, yet uncluttered, low end.
Suprisingly this technique works really good for getting that low end down. When I am done with a mix I usually run another highpass filter over the whole mix around 55-60hz to eliminate a lot of frequencies that you can't really hear or feel--and aren't reproduced on most stereo systems. This low end mush can really sap a power amp and speaker of its ability to pump. Once cleaned up it is amazing how punchy your tracks will be, without any apparent loss of low end.
I do a similar thing with a low pass filter on most of the instruments as well to eliminate any extraneous high frequences. I usually start rolling off guitar around 8khz gently, the kick drum around 6khz, toms around 10khz and snare around 12khz. The only things I want to inhabit the area above 10khz are cymbals, high hats--and most importantly--the "air" of the vocals.
It is amazing how much vocals can cut thru a mix and still keeping a high sheen on the overall mix using this method. Your seperation is often enhanced as well. And you don't have to resort to awful harmonic exciters like BBE and Aphex... which are usually poorly used and can sound very sour to me.
After I have filtered my frequencies I actually begin to EQ things. Now I have a few rules of my own when it comes to using EQ that keep things under control. Once again, these are just guideline rules that I occasionally break but I have found that they are applicable for me 90% of the time:
1.) Always use a parametric EQ. Graphic EQ's are for wusses.
2.) When boosting Q must be wider (less than) than 2.
3.) When cutting Q should be narrow--from 1.5 or greater.
4.) No cut or boost may be greater than 6db +/- in any case (occasionally broken for cutting).
5.) 75% of my boosts are less than 2 db. 90% are less than 4 db of boost.
6.) Never cut more than 8db of anything unless notching out specific small frequencies.
7.) It is okay to occaionally "pile on" a wide Q boost or cut with another narrower boost/cut if you need a radical increase in that particular frequency (this makes it sound more natural and less like a resonant peak).
Okay, when I am using EQ--which I admit I do a lot of *subtle* EQing--I always aim at doing one of two things:
1.) Remove the 'bad' qualities of the sound such as rattles, hums, hiss, muddy frequency areas and so on.
2.) If there are no bad qualities that need to go, then accentuate the positive elements.
After I have taken care of those problems I then move on to actually mixing the instruments together. I always ask myself "where does this particular track live?" and aim towards cutting other tracks that intrude on that area by a few db's. The idea is to cut away parts of interfering signals to allow certain instruments to shine in particular bandwidths. This is my general schema (these are relative and only guidelines--individual mixes/use may vary):
80hz - rumble of the bass
100hz - thump of the kick
200hz - bottom of the guitar
250hz - warmth of the vocal
350hz - bang of the snare
400hz - body of the bass
500hz - clang of the high hat
600hz - clang of the cymbals
800hz - ping of ride cymbal
1000hz - meat of the guitar
1200hz - body of the snare
1400hz - meat of the vocal
1600hz - snap of the kick/plectrum on guitar (attack)
2500hz - wires and snap of snare
3000hz - presence of the vocal
4000hz - ring of ride cymbal/top end of bass guitar
6000hz - sizzle of the high hat
7000hz - sizzle of the cymbals
8000hz - top end of the kick
9000hz - brightness on snare and cymbals
10000hz - brightness on vocal
12000hz - air on vocal
14000hz - air on cymbals
Generally I want each listed element to be the "star" of that particular frequency range--anything that is near that range that is stealing the thunder of the instrument gets a gentle 1-3db cut across a fairly wide bandwidth. For example, almost universally you have to cut guitar at 3khz to make room for the vocal--especially at high gain settings with tons of harmonics. Lower the guitar a bit in that region and POP... the vocals come out.
I realize my method is a LONG one that takes some time, but results in superior mixes for me. I like to feel that the entire frequency spectrum is represented by something unique in each area to allow the full instrumentation to shine through. I also make ample use of panning to get clarity and seperation and sometimes take that into consideration--especially when two elements are in the same frequency band. It is good to have one or both panned differently from one another. A perfect example is the ride cymbal and top end of the bass: the bass will be coming at you down the center and the ride cymbal should be off a ways R or L--thus avoiding conflict.
Hopefully this helps. I didn't give away too many of my good secrets.