I have a waveform analysis tool that provides statistics like peak, avg RMS, etc.  It is something I got back when I had Win95.  Now I think I see:  keep the final mix hovering around -12 and then after mixing hovering around -10 with peaks getting chopped off at -0.3.  That was what I was doing previous to getting all confused with this K-business.
 
For final EQ I try to "match" the EQ spectrum on Release Me by Agnes out of Europe, which hit #1 on Billboard Hot Dance music.  I like the sound of that, and use that as a guide, even though my personal stuff is vocal driven guitar oriented rock.
 
Also, I just put the SPL meter on the desk and if it goes over 80, I turn the mixing desk fader down.  I'm deaf enough.

GregGraves
My master comes out with mean RMS of 16 with a peak of 1.5, but for example a jazz CD by Hardcastle comes out mean RMS 13 peak of 0.  Other examples are even more squared.
 
Am I shooting myself in the competitive-face trying to be a Katz-ian? 

Not necessarily.
 
Many audiophile releases have average RMS values between -16 and -20 dB. Most of my favorite commercial recordings fall into that range. Most things mastered prior to ~2000 will be in that neighborhood. It's only been recently that -20 dB came to be considered "quiet". It used to be the norm.
 
Your record played on the radio will sound just as loud as the -13 dB record, and perhaps sound a little better. Your song played on an MP3 player with automatic gain normalization will also sound just as loud, and very likely be noticeably better-sounding. Someone listening to it on a CD in the car will likely notice no difference. Only if they're listening to a CD on a nice home stereo would they maybe need to turn the volume up a bit.
 
If you feel like you're falling short of commercial products, make a version that's a couple dB hotter and do a side-by-side comparison. Make sure you adjust the volume control so that you're hearing both at the same volume (radios and MP3 players and some car stereos do this automatically). If they sound equally good, use the louder master. But if the louder master doesn't sound as clear or as exciting, go back to the quieter version. In my experience, -16 dB ave. RMS typically sounds better than -13 dB, but it depends a LOT on the mix.

After reading all the above posts (and thank you), I've googled all about, and come back to re-reading the above.  My distillation is:
 
1.  Don't fall for the "louder is better" psychoacoustic trap in attempting to Master your own stuff.
 
2.  Pick a standard for yourself regarding how hot your going to mix (K-14 or K-12) and how hot you are going to master (K-12 or K-10).  Or pick some other level.  The former should be lower than the latter.  Just don't pick K=0.  {NOTE:  K-12 is same as setting your Sonar meter to Peak+RMS and keeping the RMS display around -12 db.}
 
3.  Calibrate your monitors INDIVIDUALLY with the -20db pink noise.  This allowed me to determine that my right monitor is -2.6db quieter than my left monitor at my listening position.  Not good, but good to know.  I can fix.
 
4.  Monitor calibration also provides a repeatable loudness SPL that you can accurately return to ... where the frequency response of your ears is flatter.   Use this level when mixing ... not tracking ... so not all the time.  For my room, according to the SOS article, 76db SPL.

Good summation, Greg. You might want to amend the original post to say "lazy readers can skip to the bottom".

GregGraves
 
3.  Calibrate your monitors INDIVIDUALLY with the -20db pink noise.  This allowed me to determine that my right monitor is -2.6db quieter than my left monitor at my listening position.  Not good, but good to know.  I can fix.
 

Some here will probably wish to disagree with this but others will readily endorse the suggestion:
 
3.5.  Purchase, install and configure IK Multimedia ARC to fix problem 3.

I don't think you need to buy ARC to balance your monitors well.  You have all the tools to do that accurately now. The positions of the variable pots on the rear of active monitors are not that accurate either.  Unless they are up full perhaps and you fine tune levels going to your speakers.
 
I am lucky in that my room sounds great without ARC and personally I am a believer that if it does then you don't need it.  I have checked the response of my room here and it is very flat.  I am in a larger space which has worked out rather well.  If you have a difficult space then ARC might be a saviour.  I would prefer a hardware version of ARC as I would not want the issue of running your studio monitoring signal flow in software to hear it.  It should be permanently on and have the option of turning it off.  I think KRK make a hardware type setup.
 
Mastering to around -16 dB FS is a nice thing to do.  If I were aiming at that then I would start at K-20 for the production phase and come up 4 dB rms in mastering. That is a better way to do it I think.  Just make sure the listener has got 'Sound Check' switched ON in iTunes.
 
It is still easy to say master at -16 but still many of you are not in the business full time mastering stuff as I often get into with paying clients. They still now don't like low rms levels of mastering and still insist on it being pretty loud at times.  I know we have to educate them yet.
 
Anything mastered at -14 or lower is just going to sound great. Especially when you crank that ref level up to your listening level. eg 85 dB SPL.  It means things can jump up rms wise to 99 dB SPL. Transients can be that high all the time which sound very good.
 
Many of you I feel underestimate the benefits of monitoring loud enough I think. Many of you are in no position to monitor loud for any period of time.  I am and I do it as well.  95 dB SPL is excellent and so is 105 dB SPL too.  (not for too long might I add)  There is still much in a mix you just don't hear until you really turn it up loud.  Listening down at 80 dB or under won't reveal it either.  I have solved mix problems by identifying things up loud.  I also love very low level listening too in mono through a small speaker as we know.  That is another world.
 
I like 85 dB and think it's a nice level for sure for me in my sized room anyway.  First thing in the morning though after a good night’s sleep 75 dB can sound loud for sure.  As the day wears on you can turn it up slowly over time.  But it is good to keep it consistent though over a long period. That is where a permanent SPL meter comes into its own.  It really helps you maintain that.

Jeff Evans
 
Anything mastered at -14 or lower is just going to sound great. Especially when you crank that ref level up to your listening level. eg 85 dB SPL.  It means things can jump up rms wise to 99 dB SPL. Transients can be that high all the time which sound very good.
 
 

After reading this, I went and dragged a few songs into SPlat and started looking at the transients.
 
I started off with a song by Trapeez called "Black Cloud." I have always loved this song and the sound has always "stirred" me, you know, just got to me. Listening to it now I see that it's a very open/natural sound. (Side note: this album was mixed by Roger Quested!)
 
Next song was "Lucky One" by Alison Krauss. I have always liked, was impressed by this recording. It also has a very open/natural sound to it. (Side note: This album is on Bob Katz list of albums)
 
Next was Jamey Johnson "Mary Go Round" Great album all the way. I really love Jamey's songwriting skill and his delivery. However, I did notice that this music was more compressed than the other two.
 
Next up was 3 Doors Down "Kryptonite." Another great song. But much more compressed.
 
Next was Joe Satriani "Cool #9" Love Joe's style. But to me his lead on this recording wasn't standing out. Not as compressed as "Kryptonite" but still not a lot of openness.
 
Finally "Van Nuys" by SIXX AM. This song sounds "intimate" in the quiet parts, but not as open when it starts rocking.
 
What I saw was that the more compression was used, the less I liked the sound. So when Jeff talked about the transients, I had to go looking to satisfy myself. I like the openness of "full" transients. I think we all do.
 
I am also getting to where I listen to music thinking "They shouldn't have used so much compression on that song!" I can't listen to music anymore without trying to figure out how much compression was used!! I can't just "listen" anymore!! Ugh!
 
 
 
 
 

For me I think some of it relates to the fact I have played drums live and have done so for a very long time.  Also I play Sonor drums which have a fantastic transient response.   I am hearing transients in real life all the time.   Even when recording and processing drums it is so easy to effect the transients of the drum hits so easily.   Same in mastering too.   One of the most important settings for me is the  'Attack' setting in the mastering compressor.  Too fast and the transients in the music are gone.   Any compressor with too fast Attack settings over anything will kill the transients too.
 
Also if you are striving for a very loud mastered rms level then there is not going to much room for transients above that.  But with rms levels at -16 or so they will remain alive and well. 
 
There is at least a standard now in producing music for TV. The spec says the mastered track needs to be -23 dB FS.

As I sit hear listening to my wife play her piano, I can hear the difference in the transients between the percussive nature of the higher notes as opposed to the "swelling" of the lower notes. Details in music like this I had never noticed before.
 
Why did this loudness war start?

Grem
Why did this loudness war start?

Music with no dynamic range has the advantage of being listenable in noisy environments, like in my car with studded winter tires.  Or low-level background music in a store.  You can make out all the parts because they're all at the same level.  So I see such music as a positive thing in certain circumstances (cheap earbuds is probably another).  I don't really get the loudness war debate since I just adjust the volume on whatever I'm listening through.