I personally find graphs with 3 axes nearly impossible to interpret. I would much prefer to graph using something else to represent the third dimension, but there are various choices depending on need. Color, which you mention, works well when the third dimension has a small number of discrete categories (but be aware that quite a few people are color blind; for more on using color, see ColorBrewer). Lattice graphs (e.g. the
lattice package in
R) can work very well for 3 or more dimensions.
R package, offers faceting.
I don't know, then, if there are "general" rules - I think it depends on what sort of data you have (in each dimension) and what you want to emphasize in the plots.