To elaborate a little on Baccheus' answer, comparing absolute dollar amounts is tricky. If you work for a university or private research institute, administrative overhead ranges from 25-50% (or more). So, it you apply for and win a $100,000 grant, you actually start with between $50,000 - $75,000. Then your department, division, etc. will take an additional administrative percentage.
If you want C-14 dates, it's several hundred per sample. If you want tree-ring dates it's $30/sample (minimum charge is $300).
If you need to collect field samples, all travel and associate costs are paid for out of the grant. Many facilities charge for computer time and other analytic equipment use.
Also, like a for-profit business, the salaries of research technicians and support staff at public institutions can be "billable". That means you have to cover not only their actual take-home salary, but also all of their fringe benefit costs. As I recall, the last time I did that the calculation was: salary + (salary * .25) + (.15 * (salary *.25)). It adds up.
Publishing in a peer review journal also costs money. Color illustrations can cost $1,000 / page. And although the author keeps the copyright, he or she does not have publication rights. As author, you generally get a few free copies of the article, but if you want more you have to pay for them like everyone else.
And that is the fun part.
Here is the US National Science Foundation Grant Proposal Guide:
http://www.nsf.gov/pubs/policydocs/pappguide/nsf11001/gpgprint.pdf
It is 71 pages long, and if you do not follow every single one of their friggin' rules (including everything down to the type and size of font you can use) - exactly, they can just throw your proposal in the trash and not even bother to inform you.
It is far from the "sweet deal" a lot of people think that it is.