NSF just released their funding stats for 2008 (now that it's almost 2010) and throughout the document they make comparisons back to 2001. If you are applying to NSF anytime soon the document is worth looking over, but there are a few trends that jump out right away.
1) It should be no surprise to anyone that overall funding rates are down from 2001, but only by 6%. The total successful funding rate in 2008 was 25% for all PI's, with women being 27% successful and a rate of 25% for men. In a stat that hits close to home, "new" PIs only had a 19% success rate. In this case, "new" refers to any investigator who has not previously had an NSF grant as a PI (meaning that Doctoral Improvement grants and Postdoc fellowships don't count). I don't know whether this is due to experience or something more along the lines of the issues at NIH that Drugmonkey has been discussing, but it is certainly not receiving the kind of attention that the similar NIH phenomenon has been.
2) There is also a large disparity between NSF's Top 100 (the 100 most funded PhD-granting programs) and everyone else. Among researchers at a Top 100 university, the success rate was 27%, whereas it was 18% everywhere else. Obviously the confounding factor here is that one can argue that the people with the most successful or promising research will end up at a Top 100 university and also be more competitive for funding, but I thought it was an interesting observation.
3) The number of funded multi-investigator grants has increased marginally, but the number of funded single-investigator proposals has dropped by nearly 25% between 2001 and 2008. One might assume that the focus on collaborative research might push the multi-investigator grants up at the expense of individual grants, but this doesn't seem to be the case unless the number of PIs per grant has gone up from two to >2, but those data are not provided.
4) One thing that I found very surprising was the number of grants per PI. In the fiscal years 2006-08, 83% of PIs had one grant! That left 13% with two grants, 3% with three and 1% with four or more. As someone who has three very different projects that they are trying to get funded and has been talking with a colleague about a fourth, I'm not sure what to make of those numbers. If nothing else, it concerns me that reviewers might balk at a proposal just based on the number of grants held by a PI. Now, I'm not thinking that everything I am submitting will eventually get funded in the next year, but if these projects weren't fundable in my opinion, I wouldn't be wasting my time. Perhaps people who have served on a couple of panels (*cough* Odyssey? *cough*) might shed some light on the perception of when someone has "too much" funding.
5) Anyone on the postdoc market also won't be surprised to hear that whereas the number of senior personnel and grad students supported has gone up by 52% and 23%, respectively, the number of postdoc positions supported on grants has actually dropped by 10% since 2001! This stat sucks for a lot of reasons, but makes it clear where the squeeze is in terms of positions right now. Congrats, here's your PhD and your Dairy Queen visor!
6) Yet another stat that might be expected: it's getting harder to get grants on the first or even second try. The average number of times a proposal gets submitted before it gets funded is up from 1.8 to 2.2. Not a massive jump, but it reflects the fact that fewer grants are being funded on the first go and more are going back in for a third round.
7) Finally, good reviews are not what they used to be. The number of declined proposals that scored very well is up in a big way. In fact 1 in 4 proposals that receive an "Excellent" rating are not being funded. 57% of proposals that are rated between very good and excellent are declined and if your proposal falls in the range of good to very good, your chances of getting money are only 12.5%. So, even if your proposal is rated in the highest category there is still a 20% chance you will get back your reviews and scratch your head to figure out how the hell you are going to make improvement for the next round.
Take it for what it's worth, but it's data that you may be able to use.