A new article from CNET is titled "Science, engineering Ph.D. numbers buck downturn". The article bases this claim on the latest numbers on granting doctorates in science and engineering in the period from 2002-2003, up 2.8% from the previous period. The "bucking" here refers to the new worries about drops in enrollment in graduate programs.
What the article does not point out is the blindingly obvious problem with the claim implicit in the title. Ph.D graduation trends reflect enrollment trends from at least five and probably six years earlier, i.e from the period from 1996-1998. It will come as no surprise to anyone that this period had high enrollment, given as it came at the peak of the dot-com boom (although personally I am a tad surprised, since many people chose to drop out and join startups).
Whether or not there is a true slump in enrollment right now is an interesting debate to have. But graduation figures reflecting enrollment from many years ago (and before the dot-com bust and 9/11, and thus arguably eons ago) should not be part of any reasonable argument either for or against such a claim.