I still do things the old-fashioned way (I think it’s old-fashioned, at any rate — but makes the most sense to me). I build everything with square pixels, and then resize my animation (by nesting it in a non-square, properly sized composition, and pressing command+opt+f to fit it to the comp’s size) before rendering. The only thing you have to know is the numbers. Here’s a few:
NTSC DV:
Build it with square pixels: 640×480 OR 720×534
Render it with non-square (D1/DV 0.9) pixels: 720×480
NTSC D1:
Build it with square pixels: 648×486 OR 720×540
Render it with non-square (D1/DV 0.9) pixels: 720×486
Widescreen (16:9)
Build it with square pixels: 864×486 (or anything else with a 16:9 ratio!)
Render it with non-square (D1/DV 0.9) pixels: 720×486 (for D1) OR 720×480 (for DV)
Where there are two options in the build section, the first one should be used if you are using footage with fields, because you won’t have to do any vertical stretching to resize things, which can cause nasty problems.
If you build a 16:9 animation at 864×486 and then render it out in a DV or D1 comp (make sure you do the comand+option+f thing) you can be pretty dang confident that the squished footage has the proper dimensions.
For me, building with square pixels makes sense because my computer DISPLAYS square pixels. It also helps me to remember that there is some aspect ratio converting going on. Keeps my brain working. Sort of.
-ben