I do not see why the mix has to be -3 db. It is funny that you asked. I just posted similar question. My partner (he is old school) says- mix should be -12 db. I said- why in a world would I want that? I keep my mixes just below 00. When I master them, I use compressor and normalize if I have to. It works for my mixes, also they were always below commercial loudness. I could never get it as loud. But I dont think they would be louder if I mixed them at -3. Which is actually very simple thing to do- you mix it the way you want it and then turn the master fader down to what ever they requested.
 
We got in to this discussion because I mixed all 50 tracks of the song- each track to pick at near 00. Then I recorded a vocal track and it would not cut through the mix at 00- I would either need to let it pick or bring every other track down, which would ruin the balance. So in that sense- if I had -3 db headroom on each track, I would be able to bring vocals to 00 and it would break trough. So- good advice in multitrack situation. But when you are down to 2 tracks stereo- it should not matter- I think it is minor request- if you had it at 00, they would still be able to work with it.  
 
Now- I just switched to 24 bit- I was using 16 bit. I have no idea how I will convert the tracks to 16 for CD. I thought- sonar has that option when you bounce it to the stereo track. I am not sure if I would want to master in 16 or 24 bit- have to research it farther.

I am totally amateur at this. I have little time to do a lot of reading. So most of what I learn is by doing. I used to wonder why I had to do certain things to get the mix right. After reading this thread (don't ask me where I found all that time) I see I am doing the right things in the end. So... back to learning.

One thing I have found helpful in mixing is avoid raising faders and try to lower faders. Even if this means you lower all of them. If a track seems to low in volume don't raise the fader lower all the others. Of course after one has lowered all the faders then its OK to raise them as needed, individually. Another reason to use a Control Surface. I like to at least lower them by half. -5 to -10 on an MC
 
Start low and work your way up. 
 
 

Yeah. One trick I'm learning is to set the one track I want to be heard (usually main vocal) where I think it should be, then raise the others one by one around it so the 1st track isn't drowned out.

John
One thing I have found helpful in mixing is avoid raising faders and try to lower faders. Even if this means you lower all of them. If a track seems to low in volume don't raise the fader lower all the others. Of course after one has lowered all the faders then its OK to raise them as needed, individually. Another reason to use a Control Surface. I like to at least lower them by half. -5 to -10 on an MC
 
Start low and work your way up. 
 
 

I agree with this concept 100%.  You have infinitely more headroom going down than up.  Sometimes engineers forget that or didn't know it to begin with.

For what it's worth, -3dB peak after a full play is what I have always used when I master something here. Why do I use that particular number? I really can't tell you to be honest, but I tried it and others and felt better with -3 dB peak than anything else. The same as how I prefer -6dB peak going to disc. I tried lots of different input numbers....-6dB peak just seemed like it was the best for me and what I do.
 
When I worked with Bob Katz, -3dB was what he used as well (and what he also talks about in his book) so I was quite happy to know I at least had THAT part of it right. :) Anyone wanting to argue with a ME like Katz and his beliefs should mail him about it. :)
 
For what it's worth, I do NOT believe in up-sampling contrary to what people say regarding that. If the information is not there to begin with, I'm not adding anything helpful.
 
As for the 16/44 thing Konrad, that disturbs me. The reason being, the dithering process shouldn't be done by the client nor should it be skipped. Who's to say a client will use the right dithering to 16 bit? There are some that say you don't even need to dither. My take on that is....I've dithered because I've been trained that I'm supposed to. I do not hear a difference when dithering or just mixing a 24/48 down to 16/44 without dithering a thing. But I've heard other peoples material that had some stuff wrong with it because they didn't do something right or double dithered etc. It just shouldn't be handled by a client EVER in my opinion.
 
Would I NOT dither something? No, because the majority of my colleagues as well as those with credibility that have taught me, all said I should use dithering because it's the right thing to do. If dithering brought on artifacts that took down the quality of the audio I was working on, I'd not use it and would fight them to the death that they are out of their trees. As it stands, using it or not using it....I cannot hear a difference, so I use it because I'm supposed to and there are no artifacts. Stupid reason, I know...but I'm being honest and am not claiming to know everything.
 
When I was learning to master and Katz told me to dither, I dithered even though I couldn't tell a difference. When he told me "-3dB peak" my motto became "-3dB or don't send it to me".
 
When I was taught "do not put anything in the master bus because you are then attempting to do the ME's job...and you can ruin it" I asked clients not to add compressors and limiters to their master buses unless they truly knew how to use a compressor for glue and not as a limiter that would give me square blocks of audio.
 
When I was taught "record at 24/48 and process exactly the way the material was recorded via bit/sample, I stuck to my guns as well. I've found all of the above (whether they be right or wrong methods) to be rewarding to ALL of my clients as well as myself and my own material.
 
That said, whatever the mastering house you choose requests, is what you should ALWAYS follow. However, if you don't quite approve of something, don't be afraid to ask them why they want something a certain way. :) Good luck.
 
-Danny

Thanks very much, Danny.  I have changed my view on whom to hire for mastering based on all the great comments here.

Lynn

John
One thing I have found helpful in mixing is avoid raising faders and try to lower faders. Even if this means you lower all of them. If a track seems to low in volume don't raise the fader lower all the others. Of course after one has lowered all the faders then its OK to raise them as needed, individually. Another reason to use a Control Surface. I like to at least lower them by half. -5 to -10 on an MC
 
Start low and work your way up. 
 
 

I agree with this concept 100%.  You have infinitely more headroom going down than up.  Sometimes engineers forget that or didn't know it to begin with.

 
LOL! :-)
 
Post #8 of this thread suggests why it's "OK" to mix any way you want, why it is OK to forget misinformation, and why it is OK to simply not know misinformation to begin with.
 
 

If you look at it from the perspective of playing something... as long as a good S/N ratio is there, that system can be cranked up (if desired) and heard. The final mastered level is more to have consistency between songs (so you don't have to pounce on the volume knob). The best (annoying) example of this is commercials being jacked up assuming you are outside of the room... I cannot remember which site it was but some company site had an HP commercial come up as soon as you went there that was so loud (annoying) that I simply closed the browser. I couldn't even get past that initial "WTH" moment to even give their product a chance to be sold.

John
One thing I have found helpful in mixing is avoid raising faders and try to lower faders. Even if this means you lower all of them. If a track seems to low in volume don't raise the fader lower all the others. Of course after one has lowered all the faders then its OK to raise them as needed, individually. Another reason to use a Control Surface. I like to at least lower them by half. -5 to -10 on an MC
 
Start low and work your way up. 
 
 

John, it is very nice rule when you have 5 tracks. When you have 70 tracks, you can not lower every track just because you need to increase one track level.
 
I have to say this again: when they say, -3 Db, it should be full scale. Where do you see settings for the meter scale in Sonar? The way my meter set - it is going from 00 to -39. In that sense -3 Db means something different than full scale. How do you change Sonar to show dBFS - dB Full Scale?