Larry guessed that the effective drag per aerobraking pass would be < 1 Newton. I suspect the answer is higher by around an order of magnitude.

My reasoning is as follows:

The orbital injection burn used a 640 Newton engine burning for ~ 20 minutes. The discussion of aerobraking suggests that at least a 50% mass saving results from orbital circularisation via aerobraking rather than rocketry, and that fuel is >50% of the mass arriving at Mars.

This suggests that somewhere between 1 and 3 times the actual fuel usage would have been required if circularisation was done by rocket.

Therefore we are looking at an impulse budget (force times time) of 640*20 Newton-minutes to 3*640*20 N-minutes (nasty units, but you get the point?)

Now, we have 780 drag passes, so each one must amount to 17 to 50 Newton-minutes.

I seem to recall from MGS that the aerobraking peaked for about 4 minutes per pass, giving an effective drag of 4 to 10 Newtons.

Q.E.D.

BTW, for anyone looking for solid technical data on the instruments, this is a good document, but 1.3 meg in size:
http://mpfwww.jpl.nasa.gov/odyssey/science/pip.html2

(from the "Technical" link)

http://mpfwww.jpl.nasa.gov/technical/index.html

- Nick