It all depends in how you will use the model and on how much you can use ZBrush`s(and then the program in which you will render/compose) capacities. For me an evenly distributed mesh is enough. Like a sphere, I can make a decent head out of it, as the one in my thread.
If you can make a model in other app with more detail, more polys in certain areas and keep a good edgeflow, even better.
The best way to learn is to look at existing models, there are a few for free on the net. Except those on the Pixologic site, I like these two:
-one from Undoz on this page
http://www.zbrushcentral.com/zbc/showth ... post389451
-one from selwy
Download them and in ZBrush study them at the lowest subdivision level with the frame on.
For the second question, if you have a whole mesh it is very easy to separate it as you want in ZBrush. You`ll see on those models I was speaking about, they are separated in polygroups(different colors). That can be very useful.