I've always 'home mastered' my own tracks usually with Ozone5 or T-RackS, and FabFilter L.  I've just finished a track and got it to a loudness level that doesn't cause artifacts, etc. but it's still not as loud (probably 6dB different) to a track I did a couple of weeks ago after almost identical treatment.  In fact the waveforms on their respective bounced X3 tracks look almost identical.  I was wondering why this is? I assume there's a technical explanation?  I thought tracks with very similar waveforms would sound more or less as loud as each other.
It means that if I put these tracks together on an album then I have to go back to some and re-master them downwards and make them less hot so they all match the one that sounds the loudest.

Mastering (properly) is a VERY technical / complex process.  Plugins and stuff make it easy to do with presets and stuff like that, but there are a lot of things that could factor into that.  As an example, you could have some low frequencies in one song, that are triggering the limiter (but aren't very audible, because they're either outside of the range of human hearing, or just not very noticeable) whereas, if you have higher frequencies (midrange frequencies, etc...) that are causing the limiter to trigger, then they can cause pain to your ears, because they are RIGHT THERE, hitting your ear sort of at "bullseye" or whatever.  (if that makes any sense). 
 
So even though in the end result, you might have two waveforms for the mastered tracks that look similar, if you look at them through some sort of spectral analyzer or something, you'll PROBABLY see that the "louder" tracks have more action happening in the midrange frequencies, where the quieter tracks might have more mud & stuff, happening down in the sub-bass frequencies where you can't hear them.
 
There are other things that could factor into that, but that's one thing I would suggest checking (if you haven't already).

when mastering, I always bring all mixed songs on an album into a new project. this way I can use solo buttons to compare all tracks. use you ears. also bring in a professionally mastered track in a similar style. not only can relative loudness be compared, but overall eq, vocal level, etc. a trick to use is to insert one blank track and solo it. this way, if you accidentally UN solo all tracks, you don't have to hear them all paying at once!

Don't know if this is relevant but in the older versions of Pro Audio and the first versions of Sonar there was a setting that had to be changed if your loudness was different from one version to the other. I'm sorry I can't remember specifically what it was but it seems like it was in the registry somewhere. After Sonar 3 there didn't seem to be that issue anymore but I do remember having to set it in earlier versions. Maybe someone with a better memory can help out!

BlixYZ
when mastering, I always bring all mixed songs on an album into a new project. this way I can use solo buttons to compare all tracks. use you ears. also bring in a professionally mastered track in a similar style. not only can relative loudness be compared, but overall eq, vocal level, etc. a trick to use is to insert one blank track and solo it. this way, if you accidentally UN solo all tracks, you don't have to hear them all paying at once!

A better idea is simply make use of the Exclusive Solo button when auditioning tracks

Thanks guys.  A lot of excellent info here.
I put the CW analyser on the un-mastered track, i.e. the mixdown, see below.  It's saying there's too much energy in the bottom.  (What shape should I be going for?) It's an up tempo song and I really wanted kick and bass to drive things along, but I've now got to arrive at a happy medium I guess, whereby there's drive and thump, but not so much that I'd like, but the overall track can be compressed better.
 

If you can find a song that has a similar bottom end as what you're trying to achieve, I'd bring that in and run Analyst on that for comparison.   The differences could be subtle but just enough to cause compression/limiting to have to work harder.

And that stuff can really get tricky once you start digging into it... Certain things like harmonics can drastically affect how your audio sounds.   You might not want to "match" whatever the other bass-heavy songs LOOK like in the analyzer... Think of it as when you're at a concert, and you might FEEL the kick drum knocking your throat through your spine, but you might not be able to HEAR it very well necessarily... FEELING that thick kick sound, is what the low frequencies are going to do for you... But the harmonic frequencies above that kick's fundamental frequency, are the ones that will allow you to start HEARING the kick along with it... Sort of gets complicated if you've never dealt with that sort of thing before... But yeah, anytime you see a DRASTIC curve like that, then it's a safe bet that those low frequencies (that you can FEEL but can't HEAR) are going to trigger your limiter, before the audible high-mid frequencies do, and therefore, your song is going to SOUND quieter... (but it might make the subwoofer in your vehicle shake like crazy).  But realistically, as a starting point, everything below 80Hz you PROBABLY can't actually hear very well... so there is NO reason that the peak point of your song's frequency spectrum or whatever, should be down below 80Hz.  (Just my opinion)

Thanks Etmos.  I appreciate this sort of detail.

Lots of good info on this thread.

skyline - you mentioned having Ozone 5?
...any chance you have their metering plugin "Insight"?

I personally think that the spectrogram, as opposed to the spectrum analyzer, is often times easier to use to
see trouble frequencies that you cannot hear.
(It helps you to see "build-ups" of frequencies in those low or low mid regions.)

Voxengo's SPAN is one of the best free metering suites, and includes a great spectrogram.

There are tons of advice sites and podcasts now that talk a lot about high pass filters...
...they are your friends...

Happy Mastering,
MG