Pete Casazza discusses this briefly at the beginning of his article *A Mathematician's Survival Guide*, available [here][1]. He seems to advocate the approach of avoiding telling people what you do for a living (if you are a mathematician), and I find that I prefer that approach myself. Sometimes it can't be avoided, but in those instances I think I'll often try to move the conversation on in some of the ways suggested by other posters on this thread.


  [1]: http://www.math.missouri.edu/~pete/pdf/140-MAA.pdf