konradh
How do you know which is right?  Home and car stereos often jack up the bass, and earbuds are all over the place.  

I try to get them to sound good where they'll be listened to. I also try really hard to make sure the band think it sounds good. Really, the band members are the people I listen to most. I struggle not to force them to like what I like. I try to keep working until all of the band likes it too.
 
As far as listening environments, I try to make it sound good in every listening environment I can listen on. If my car boosts the bass, your probably does too. If my wife listens with pop eq settings, yours might too. It helps to hear the variety. I do try to make sure it sounds pretty great with the EQ flat, but I keep a balance. For some people, I might have to tell them to push up the bass. :-)

I would again urge you to check out Sonarworks... If your brand of headphones is on their list of calibrated headphones, you can download the free trial.  It is the most direct way of getting a good headphone mix that translates to the real world.
 
Sonarworks has calibrated most high quality headphones use for studio monitoring.  and as Daniel noted, they work great.  Several of us urged them to model the KRK's, and within a month they'd done it.  A great company.
 
Like I say, if you have a high quality pair, they've probably modeled it and it's a 30 day free trial of the calibration software... If you're using crap headphones, you should upgrade... get something they've modeled and calibrated and all your headphone mixing sessions will yield better results when played on speakers.  The even have a calibration service if you have some high end headphones they haven't calibrated.
 
Decent monitors and room tuning are important, costly and time consuming to achieve.  Not saying you shouldn't pursue this but headphone mixing is often convenient if you're in a home based studio and I'll just say it this way:  you are foolish not to try out the software...
 
 
 
 

I use room eq from http://www.roomeqwizard.com/ to create a sample of my listening environment.
 
I used it to create a list of corrective eq settings that I setup using Sonitus EQ (in my case 2 instances to get enough nodes). I created a room compensation eq preset by putting the 2 instances of Sonitus EQ in an FXChain preset. I use that preset on an a "monitoring" bus that is fed from the main mix bus that outputs to my monitors.
 
Then when I bounce I just bounce the output of the main mix bus.
 
 

gswitz
Really, the band members are the people I listen to most. I struggle not to force them to like what I like. I try to keep working until all of the band likes it too.

 
I don't work with outside clients any more. When I did, I thought I had a responsibility as their sound man to get the sound right. If they didn't like it I'd explain to them what I was doing and why. This led to many conflicts, as it turns out that musicians think they know best about audio engineering. In the end I would give them what they wanted, but I made a lot of records I wouldn't want to play for anybody.

Larry Jones
When I did, I thought I had a responsibility as their sound man to get the sound right. If they didn't like it I'd explain to them what I was doing and why.

It is often uncomfortable for me to stretch to the mix the band is asking for. For me, there is usually a range of mixes that are really good. Rather than going with the first one that I am happy with, I try to keep working. I can always return to listen to my first export. And I feel this has helped me be a better listener.

I have picked up techniques that really help.

One thing I have to resist is giving them a bad mix of what they are asking for. 'You want bass? I'll give you bass!'

I can't bring myself to deliver a mix I hate, but I can experiment along the way.

Forget mixing on headphones. Music is nearly NEVER heard in ideal conditions. Compare your mix to professional recordings on ALL systems, ESPECIALLY terrible ones.

a. Boominess can come from the most unexpected sources, not just bass and bass drum. Check each track.

b. Mastering tools like limiters, iZotope Oxygene, etc. can hide problems. Listen with them on and off, especially those on master busses.

c. If you can, take your DAW to the non-ideal location and listen right there to each track.
 
d. Mixes should be tested, yes, with very good monitors, but also on car hi-fis, ghetto blasters, earbuds, headphones AND in boomy, resonant rooms.


*** MY OWN PAINFUL JOURNEY OF DISCOVERY ***

I mix on Yamaha HS8 monitors in a well-damped 3x4m room. (I also switch between cheap computer speakers, Neumann KH120, Adam A5X, a butchered mono transistor radio, the matching Yamaha HS8S sub and Logitech gaming speakers with their own subwoofer - just to make that clear.)

A client complained my mixes were too bass-heavy. I disagreed. I went to his home and he has the SAME SPEAKERS I USE (HS8s) in a small echoey study with a large window down one side. One speaker is on a metal shelf, the other behind the door.

He played my mix and yes, it is boomy and bass-heavy.

I protested the poor acoustic conditions. He put on some commercial rock and pop CDs and perplexingly they sound good! (by comparison). He looked at me with a look that said, "You are clueless".

I moved around his study and found areas where the booming is worse, but I had to agree, my mix suffers far more. Back in my studio everything seems OK. I have no particularly boomy areas in my room.

I asked if I can bring my laptop and interface to his study for a test and he agreed.

What I discovered is most amazing:

1. The electric distorted rhythm guitar has a bass component. Switched to solo, it sounds warm and powerful - just the way one would want. But it had a resonant frequency that coincided with one in the client's boomy study. I had to crank HP rolloff all the way up to 250Hz to get rid of it. I found out that in the total mix this resonance hit the compressor and caused the bass guitar to duck. In my studio, all I'd noticed was a tonal shift. No boominess.

2. The bass guitar (recorded DI and mic'd) had a note where only the fundamental was present, hardly any harmonics. This made that note sound weaker so the bass player unconciously played it louder. Again, this found a boom frequency in the client's study, but seemed fine in my studio.

3. Finally, the floor tom also boomed. I'd shortened the decay with an envelope shaper which sounded fine in my studio, but again, it picked up a boom frequency at the client's room, so suddenly it rang longer than I'd shaped it to.

Fixes?

1. With the guitar rolled off so extremely it sounded thin on its own, but it worked in the total mix. So I left it like that.
In fact, it was here that I learned, like MondoArt has stated, that it is good practice to trim away frequencies from instruments that encroach on those of others, especially in the lower frequencies, but also higher frequencies, because they can also mess with mastering tools in unexpected ways.

2. Bass guitar was not fixable. I personally replayed all bass tracks, taking care to hear out the fingering positions of the original bass player and plucking nearer the bridge to create more harmonics and always watching the audio as I recorded to avoid peaks.

3. I replaced the toms manually with samples. Fortunately tom rolls only happened occasionally so it was only a day's work. Today I would just use the drum replacer.

*** RESPECT ***

I have enormous respect for mixing engineers. It is a huge amount of work with lots of gotchas and little thanks.

I also understand why the bands get thrown out when the final mix is done. The talent would be horrified to discover what really happens in there.

jpetersen
In fact, it was here that I learned, like MondoArt has stated, that it is good practice to trim away frequencies from instruments that encroach on those of others, especially in the lower frequencies, but also higher frequencies, because they can also mess with mastering tools in unexpected ways.

Any time I'm tempted to hi-pass something past 100-150Hz, I check myself, and use a low shelf instead. I think everyone at some point hears the advice to high pass everything except the kick and the bass to create this huge space for them, and it sounds like the magic you've been waiting for, especially if you've been struggling with mud and boominess in the low end. So they go nuts filtering everything at 200Hz and higher, which admittedly leaves a lot of room. But it also sucks a lot of warmth out of the mix and gives an overall brittle sound. I find that you can clear a lot of mud with a shelf cutting a few dB up to 250Hz or so, whilst retaining some of the warmth. I almost never hi-pass any higher than 80-100Hz on an instrument, and I've heard a few people say they don't do it at all. 

sharke
Any time I'm tempted to hi-pass something past 100-150Hz, I check myself, and use a low shelf instead.

I use low shelf often on my dreadnought guitar.

Similarly, rather than completely muting track, I tend to dim it.

jpetersen
In fact, it was here that I learned, like MondoArt has stated, that it is good practice to trim away frequencies from instruments that encroach on those of others, especially in the lower frequencies, but also higher frequencies, because they can also mess with mastering tools in unexpected ways.

 
I sort of agree, but you have to make value judgments, and choose what will produce the desired results; also sometimes it's the boosting, not the cutting, that causes the differentiation. For example, suppose kick and bass are fighting, you want to keep them centered in terms of stereo position, but you want the bass more prominent. You have several options.
 
1. Mix the bass higher.
2. Mix the kick lower.
3. Use EQ for a little less low end in the kick.
4. Use EQ to bring up pick noises and harmonics in the bass, then trim the bass's low end a bit to let the kick provide a greater proportion of low frequencies (which will overlap with the bass).
5. Slide the bass just a tiny bit (a couple milliseconds) ahead of the kick.
 
Also remember that a mix is an interlocking series of frequency spectra from the various instruments. For example, you may want to dip the midrange a bit on rhythm guitar when the singer is singing.
 
For this reason I start off mixes in mono so the instruments will "collide" with each other as much as possible. Then, you can adjust EQ to emphasize or de-emphasize particular instruments. Once everything sounds distinct and has its own space in mono, when you start creating a stereo field the whole mix widens up and creates a full, well-defined space. It's also more translatable, because the playback system's frequency response becomes less crucial as you've already done frequency response tailoring. You're not counting on the playback system to be capable of nuanced response.

I'm trialing the Sonarworks plugin now. I opened the profile for my ATH M50's and the problem I have is that the plugin seems to attenuate the overall gain by a significant amount. Everything sounds quiet through it, even when I hit the big "bypass" button. If I turn the plugin off completely, the normal gain level resumes. I'm not sure what I'm doing wrong. There is a level meter with a slider and that's turned up as far as it will go.