I usually work to around -10 LUFS. This one had a few quiet sections in it, so it came out around -11LUFS, but the final choruses are hitting pretty loud - usually around -8 or -7dB RMS. Dynamic arrangements tend to be the key in finding the balance between loudness and not getting an overly squashed-sounding mix.
That said, a big part of achieving healthy levels and the kind of "density" that modern mixes have is fairly stringent dynamic control before the mix even hits the master processors. For example: Most of the time, my mixes are hitting about -15 to 12 RMSFS at the loudest sections of the track even before any (master) compression or limiting is applied. I find, if you can retain good transient response with that kind of density, it bodes well for when you apply limiters and the like.
Yeah, I'm lazy that way - once I get a collection of songs, I'll take the unmastered files and master them together, but just for individual mixes, I throw some master eq and limiting on the whole shebang. I used to master each song in a separate session, but once I got a faster computer with more CPU, I got lazy. I still think giving the mix a rest and mastering the song in a new session the next day is a far more objective approach that yields better results. I just get impatient and lazy when I finally finish a mix!