Personally, I don't think the Math GRE really says anything about a student.
"How well can you work tricky Calculus/Pre-calculus questions with a time limit?"
What that has to do with someone wanting to study analysis or algebraic geometry or topology is beyond me. Personally, I'm an aspiring complex analyst. How much is my knowledge of complex analysis going to help on the exam? Three questions at best.
I dunno, I just think its silly. Seems the test is more for people who sit around thinking about Calculus, but in actuality, wouldn't that be the engineers and physicists? Don't most serious mathematicians move on to algebra and/or analysis and/or geometry (and/or a combination of the above)? I know that if not for this test I wouldn't be thinking about computing things in my mathematical thoughts.
(I'd be spending 100% of my time studying Rudin, Ahlfors, Markushevich, and Birkhoff/Mac Lane!!!!)
Of course, even more generally, not all mathematicians are solid at the same topics. One of my professors had a classmate who was awful at algebra, but was a genius at topology/analysis, and actually got a very nice offer from a top 10 school. So again, maybe someone just isn't good at calculus or algebra, but is a freaking genius with analysis or topology. Their score probably wouldn't be so hot. But maybe they're the person who will prove Riemann's hypothesis!
Bah! Just complaining because the test looms ahead for me, and is freaking me out.
And while we're on the topic: Anybody have any clue what good ole' Wash U in St. Louis looks for on the Math GRE?