I'll play devil's advocate to the comments above, and suggest that it's not necessary to do a 2-step process but rather is perfectly acceptable to master in the project. If it's part of a collection (e.g. a CD or a film cue), or if you're sending it out to a third party for mastering, then that's another story. But for a one-off standalone song, I say do it in the project.
 
lick4lick has a valid point about removing the temptation to mess with the mix, but I'm going to suggest there's nothing wrong with that. After all, a lot of the things MEs do to improve the final mix are actually things they wish they could do in the mix. That's one reason they like to get stems, so they have more mix control prior to mastering. Sometimes, they'll even send it back to the mixer with suggestions for how it can be improved so as to result in a better master.
 
Truth is, you will find things out about the mix once you start mastering, deficiencies that were not apparent until you glued them all together in mastering. If you use any bus compression on the master it will change your mix, sometimes by a lot. If you find you're using a lot of EQ on the master it could be an indication of spectral balance issues that would be better addressed in the mix. Other tricks such as expanders and transient shapers shouldn't be necessary at mastering time; if you find that they do improve the mix, they'll work even better if you figure out which individual tracks need expansion and apply those effects at the track level.
 
MEs will tell you that it's possible to create a mix that needs nothing in the mastering phase except volume adjustment. Few of us are that good, though, and the rest of us have to bounce back and forth between the master and the mix. I know when I'm headed in the right direction when I have to do less and less to the master.  I couldn't do that without having the mastering chain inserted right there in the project.
 
Not saying any of the previous suggestions are wrong, just my $0.02.

Just to show how complex things could get and to illustrate how much there is to keep-track-of when Mastering, I made this video.  There's potential for infamous SNAFU scenario when I Mix and Master on the same Sonar Project ...  Point is well taken bitflipper.  I had to re-mix numerous times, then bounce, then export, then import  ... a whole lot of work.  But, I'm afraid I'd mess things up worse if I don't separate.  I've got this work-flow that seems to work.  I thank everybody for the input ... great to hear other opinions.
 
Anyhow, here you go ...
 
https://www.youtube.com/watch?v=8fHgjMfjE3E
 
 

They say it is good to get a mastering engineer to master your track and one of the main reasons is they have never heard it before and hence it is fresh to them.  They are very unbiased.  If however you are mastering your own material then the closest thing to this is to mix the track and let a week go by before mastering it.
 
For me the main reason is you have assaulted your ears for many hours doing the mix so I find I cannot make great mastering decisions straight after that.  Any EQ setting you do on the mix day will be different to an overall EQ decision you make a week later.
 
I prefer to get the mix sounding as good as it can first without mastering.  Another reason too is it is nice to have a full premastered version backed up too which you won't get when you master all at once.  It is nice to listen to the mix as well for a week.  In the car, everywhere.  You will hear things and make slight changes as a result.  Hence a better mix.  They say mastering into a compressor changes things so that is fine put it on your master buss and apply some gentle conditioning there.  But remove it before printing the Presmastered mix though.
 
Wait a week and create a proper mastering session and you will find you make better mastering decisions later on.  You will end up using less of everything when you do it this way.  This is only good if time is on your side though.  I recently had to mix and master at the same time in order to belt it all out fast so yes I did that.  I found though I only needed some compression and limiting to bring the overall level up and it worked fine but for me it is not the ideal method

Jeff ... glad you mentioned letting time go by prior to Mastering your own Mix ... really important to recognize "ear fatigue".

lick4lick
Jeff ... glad you mentioned letting time go by prior to Mastering your own Mix ... really important to recognize "ear fatigue".

Exactly. On the mix day you will find you are still boosting highs and adding 3 or 4 db at the top end.  But the only reason you are doing this is because your ears are shot.  A week later (after a good nights sleep and you are totally fresh in the morning) you will not only realise that your mix is too toppy but you will find yourself pulling the highs down or rolling them off.  Quite different scenarios.
 
I have read many articles where even the best engineers in the world have said they really regretted mastering so close to mix in order to satisfy the record label deadline.
 
You will always get a better master when you separate the two. Having a totally premastered version also allows you to remaster.  As it sometimes takes more than one attempt at mastering to get it perfect.

Jeff Evans
Exactly. On the mix day you will find you are still boosting highs and adding 3 or 4 db at the top end.  But the only reason you are doing this is because your ears are shot.  A week later (after a good nights sleep and you are totally fresh in the morning) you will not only realise that your mix is too toppy but you will find yourself pulling the highs down or rolling them off.  Quite different scenarios.
 

 
Funny. I was just getting ready to boost the highs on a mix this afternoon when I decided to quit for the day. We'll see how I feel about it when I go back to it.

Yeah I get that too.  And in a few days if you still feel like boosting the highs then yes the mix may need it for sure.  But then again you may have a totally different view and find you are actually easing them down instead! A lot of mixes are way too toppy.  A good mix wont be actually but only have a few elements in there that are clear and have nice top end.  Put a spectrum analyser over your mix.  The best mixes I have heard all have a gentle slope downward from about 5 K onwards.

Great tune, btw. "I feel like a coyote on the freeway". Love that simile.

I do a two-step process inside the project.
I bounce 2-3 slightly different stereo mixes, and archive the rest of the tracks when I start "mastering".
That "slightly different" means usually mixes with different EQs or levels on instruments I usually have difficulties with - hihats, cymbals and the very low end.

I mix straight into my mastering chain so I know EXACTLY how it's going to sound when it gets to the outside world.
 
But I also have a Pre Master bus which I use for exporting individual songs which are then re-imported into my album project with the same mastering chain.
 
Mixing with the K system in mind and calibrated monitors means that I have very little overall level differences between songs