[megatherium]

Lots of folks (US nationals) got math and physics PhDs in the late 80s/early 90s and discovered that the academic job market was horrible. This came after the universities whined about an impending shortage of scientists and engineers, said to be caused by the low numbers of Americans who study science. The universities got what they were really after: increased funding for research, including graduate programs. This greatly increased the number of graduate students. For example, in mathematics, the number of PhDs awarded went from less than 800 in 1986 to more than 1200 per year in the early 1990s. Half of those PhDs were awarded to foreign nationals, who helped glut the job market. If Americans aren't entering scientific careers after that, it's no wonder.

[laurav]

The academic job market is wretched, but there's no need to go into it if you have a math/science PhD (unlike, say, a PhD in "communications studies"). The government hires math PhD's as code crackers and for other national security stuff. Pharma companies hire science PhD's as researchers -- and it certainly looks like the drug industry will have the funds to keep that market brisk for a while. Then there's aerospace or general high-tech companies, robotics, etc.
I'd blame the fall in people getting these degrees more on the lack of challenge in K-12 than the job market.