ANDS, what are you on about? Most departments have GRE cutoffs on applications they consider. The only substitutes I can think of for a low score is a strong showing on the Putnam, or a recommendation letter from a well-known mathematician.
Define "most", because your most and my most aren't living in the same neighborhood. I've spoken to several people who have been admitted to top tier schools (UNC, Harvard and Chi-Town for starters), and they didn't knock their subject into the 80th percentiles; so either grad commitee's have quite a bit of latitude in reviewing applications or these people were lying about their scores.
Otherwise one shouldn't waste his or her time applying to the top schools because the outcome is clear.
Absolutely ludicrous. I wonder how many wasted opportunities people have left on the side of the road because they have bought into this idea that performance on the subject GRE absolutely defines ones perparedness for a math PhD program. Which is silly. Now if this fellow was getting B's and B+'s, with the occassional A - then his chances would obviously drop, and a low score is going to be a good predictor of success in the graduate program.
However, he says he has mantained a good A average throughout his upper division career, and depending on the school that is nothing to sneeze at. About the only thing I can think that would sink him/her is whether or not they've done any math related extra-curricular activities, and how strong their LOR's are going to be.