I use a few different work-flows, but I haven't really found one I like completely which is frustrating.
My favorite if I can make it work is to use T-Racks 3 for the mastering and trimming. I really like the interface there. You can set-up one chain, but change it or have different settings for different songs. It is easy to have a reference there and bounce between songs to check for consistency between tracks and such. It also also has a very good metering section, A/B with volume adjustment/matching, batch rendering, and does basic trimming and fading. Definitely more than enough for online distribution or something like that. And the plugins are pretty decent. BUT... of course you are stuck to T-Racks plugs and you get no CD capabilities at all.
If things go well with T-Racks, then I do the CD-Layout and PQ codes in CD-Architect. Quick and easy. So I will do this for like Demos, or one off songs that I am not charging a lot for. CD-Architect is very capable, but only supports DirectX plugs.
The problem is that I now have external hardware, and other plugs that I like better that I will use in combination with the T-Racks plugs. So I bought Wavelab 6 with the crossgrade from Cubase. It is a pretty clunky program you can tell it was designed to be more of a stereo editor. The master plugin bus is cool, but it doesn't save in a session, you have to save it as a preset. Also if you are doing a montage, you are stuck with it over the whole session.
So for stuff like albums I take two approaches depending on the consistency of mixes and the songs. But every band has the one ballad or the one rocker or the one that sticks out among the rest.
One is to master each song individually using a similar master chain, then do the cd layout separate all in wavelab. Then use the built-in RMS matching or just do the best I can A/B'ing to get it all consistent. Compilations and such that are already "mastered" I will do this way too. Or if the songs are just drastically different, or mixed by different studios, or inconsistent mixes where each one requires a different level of saving... etc.
If things are pretty consistent and sounding pretty good overall then I will bring them all into a Montage with the layout and load up the master bus. But then I will put additional plugs onto each track in the mantage if needed, usually only one or two, or maybe a touch of EQ. Do fades and curves and use volume adjustments on the Montage to get it all consistent or for the album to flow. Then I can export digital files, the whole montage for review, and then burn the CD with CD-Text, ISRC's, PQ's, etc all in one shot. But the mangement of this is a pain since the master-bus is a preset that you have to remember to save and the montage is separate, you can save it, but it depends on the original files, so you can't "archive it". Also A/Bing with a reference is more difficult.
I have also done the whole load up each audio file as a separate track in Cubase and then work accordingly. It is great for A/Bing and such, but fading into other songs and such and listening to how the tracks flow together is difficult and I think an essential part. Also the mixdowns are difficult and take a long time since you have to do each track one at a time if you use any plugs on the master bus. To me I found the T-Racks or Wavelab way faster since I can start the render and walk away. So I don't do it this way.
So that is how I do it, and they all suck workflow wise. But the job gets done and the clients are happy. It just sucks when you are trying to focus on the sonics but then distracted and bogged down by all the workflow suckiness. So I am all ears to see how other guys do it and maintain the good sonics, song to song flow, relative consistency (minus the "hits" being a bit louder and such), and a reference.
The reference I could probably maybe go without. It is just that some folks like things dark and doomy, or bright and big, or brown and punchy, midrangy, midrangeless, harsh and clipped, etc. So I try to get things close.
