When mixing a band, I usually compress individual instruments to level out the peaks.. there's a nice compression in Goldwave "reduce peaks" you can listen to it, adjust treshold so that you first hear it squashing then back off until the squashing stops-->press ok, and then you can see it leveling the peaks in the wav-chart somewhat.
It helps in the end, when you put them together, there's no huge peaks in the final wav-chart, that start clipping if you increase overall volume.
Ofcourse, if you can compress the peaks allready on recording, using gear or adjusting playing it's better...
Finally, when the mix is done, I usually normalize it to 0db using wavelab --> reduce peaks in goldwave once more, then add gain or normalize again in wavelab. Just to get some volume.
Ofcourse, if the final mix is low enough to begin with, I try out wavelabs tubecomp-gate-limiter OR softkneecomp-gate-limiter to pump it up to begin with. Which sounds better. Then I might dabble with multiband compressor to take out some mud in the low-middleground, and a puncher to clarify things up a bit. But with these you must be careful since they might turn some areas, like cymbals into sizzling crap...
..but for solely gain reduction, I wouldn't use a compressor...
.02