Also, you don't need to keep the entire frequency range of every instrument.  Each instrument should have a purpose in the mix.  With that in mind, cut out the frequencies that don't support that purpose.

iic Izotope has a plugin for this, Neutrino afaik.
you put it on each track and it does the eq and compression for you.
not cheap though, but it can operate in a zero latency mode as well.
 

One area where you can get caught is in the very top end. Many instruments don't need to extend anywhere near up as high. Use you LPF's to limit the top end of things. Learn to love the 48db/oct slopes on your filters too. They are there for a reason.
 
Even though a part might sound like it is only going up to say 4K or so once you put a LPF on that and set a steep slope around 4K it sounds better and cleaner.
 
With top end as well there should only be a few things that have top end. Two or three at most. The rest no. Remember if everything is bright then nothing is bright.  Think of a very dark sky with just a handful of stars in it and most of them are dim with 2 or 3 bright stars. That is how the top end of your mix should be. 

For me it all starts with sound choice. If you can hear a basic separation between all the tracks without putting a single EQ on them, you've made the right choices. Don't know how many times I've spent hours experimenting with the most insane roller coaster EQ curves trying to "carve space" for tracks which just did not work together. What a waste of time - the best you're going to get is a bunch of weird sounding tracks that still step on each other. 
 
Have you ever had the experience of just throwing a few parts together while playing around - whether loops or samples or synth sounds or whatever - and they just sound great from the get go? Excellent clarity and definition without a single EQ curve. That's what it's like choosing the right sounds, and what you should be aiming for. From there, any EQ you do is just a matter of polishing things up a little. 
 
If two guitar parts are stepping on each other, then I will dial in a completely different tone for one of them until they contrast enough. Sometimes I'll be using a snare sample which masks a lot in the mix, and rather than spend hours trying to tweak it with EQ, I'll just look for a different sample until I find one that fits nicer in the mix. 
 
As Rob says above also, arrangement is just as important. If you're clever with the arrangement so that no more than two or three things are sounding at once, you can make a lot of tracks fit together and give the illusion of a busy mix without anything stepping on each other. For instance, on the EDM forums I hear so much talk of all these insane sidechain routings and crazy amounts of EQ to get a kick and bass part to work together. And yet if you arrange your kick and bass so that they're not playing together, they both have definition before you've inserted a single plugin. 

One other point on arrangement - you can "feature" different instruments in different parts.  In other words, bring out the acoustic guitar in the chorus, feature the hammond in bridge, etc.  You don't necessarily mute the tracks you're not featuring, just pull them back in the mix and bump up the "featured" instrument.

sharke
For me it all starts with sound choice. If you can hear a basic separation between all the tracks without putting a single EQ on them, you've made the right choices. Don't know how many times I've spent hours experimenting with the most insane roller coaster EQ curves trying to "carve space" for tracks which just did not work together. What a waste of time - the best you're going to get is a bunch of weird sounding tracks that still step on each other. 
 
Have you ever had the experience of just throwing a few parts together while playing around - whether loops or samples or synth sounds or whatever - and they just sound great from the get go? Excellent clarity and definition without a single EQ curve. That's what it's like choosing the right sounds, and what you should be aiming for. From there, any EQ you do is just a matter of polishing things up a little. 
 
If two guitar parts are stepping on each other, then I will dial in a completely different tone for one of them until they contrast enough. Sometimes I'll be using a snare sample which masks a lot in the mix, and rather than spend hours trying to tweak it with EQ, I'll just look for a different sample until I find one that fits nicer in the mix. 
 
As Rob says above also, arrangement is just as important. If you're clever with the arrangement so that no more than two or three things are sounding at once, you can make a lot of tracks fit together and give the illusion of a busy mix without anything stepping on each other. For instance, on the EDM forums I hear so much talk of all these insane sidechain routings and crazy amounts of EQ to get a kick and bass part to work together. And yet if you arrange your kick and bass so that they're not playing together, they both have definition before you've inserted a single plugin. 

+1
I remember once spending hours trying to make Scarbee Rickenbacker Bass fit in my mix because I was convinced it was the highest quality sampled bass I had. I ended up dialing in a basic Fender Jazz (possibly even Cakewalk's?) and instantly everything made sense.

Hi, alot of times in a mix I'd toy with the frequency EQ, when what I was listening to, wasn't even that.. the main noticeable dynamic you'll notice before EQ is the compression, I'd do multiband compression on all tracks first before adjusting EQ, if it's your cup of tea
 
Thanks,
 
LWD

Jeff Evans
One area where you can get caught is in the very top end. Many instruments don't need to extend anywhere near up as high. Use you LPF's to limit the top end of things. Learn to love the 48db/oct slopes on your filters too. They are there for a reason.
 
Even though a part might sound like it is only going up to say 4K or so once you put a LPF on that and set a steep slope around 4K it sounds better and cleaner.
 
With top end as well there should only be a few things that have top end. Two or three at most. The rest no. Remember if everything is bright then nothing is bright.  Think of a very dark sky with just a handful of stars in it and most of them are dim with 2 or 3 bright stars. That is how the top end of your mix should be. 

THANKS FOR THIS! This has helped me a lot. Working on a very dense mix. I love these forums.

I think my biggest struggle is stopping upfront drums from masking instruments that are floating in the background. Never got the hang of that. It's something I always liked about Aphex Twin - he always has these great snappy in your face drums going in the foreground, while dreamy layers of ambience and quirky synth lines drift in and out of the background. And yet the ambient layers were always very legible with enough clarity to hear everything that's going on. I always end up having to carve frequencies out of the snare and the kick to the point where they lose their body and snap, and even then I don't get the clarity I would like. 

If eq'ing alone doesn't work you can use sidechaining to do frequency dependant ducking/compression/expansion. Also panning and use if the stereo field is important. Roll off any unused frequency space and shape the prominent frequencies of sounds. Using a frequency reference chart is a good way to do this. Cutting frequencies at 200Hz means at another harmonic like 800Hz it will be perceived to be louder.