The people making these estimates are the USGS.  With much pardon to my friends in the USGS, I do not believe them for a second.  The USGS is not engaged in finding and developing oil and gas.  Oil companies are.  (I used to work for the USGS so I know a little bit about their capabilities or lack thereof.)

A reasonable P50 estimate for ANWR might be 2 billion barrels. The structures are small and complex.  12 billion is out of the question, and 60 billion is for people who don't know what a rock is.  No one in the industry is predicting finding another Prudhoe Bay or Kuparak River field in ANWR.

And of course no one has found a drop of oil in ANWR.  There could be nothing there worth trying to go after.  If it ever opened up, the industry might drill 5 to 10 wildcats, find some small volumes, and walk away.

This is a crucial point. I recall learning in grad school that the USGS has only drilled one hole (or not many more) and that the results of the tests were not public. The probability assessments noted above seem to be based on statistcal modeling using geological features:

"An evaluation was made of each of 10 petroleum plays (similar geologic settings). For each play, USGS constructed statistical distributions of the number and size of potential accumulations based on a probabilistic range of geologic attributes."

http://www.eia.doe.gov/pub/oil_gas/petroleum/analysis_publications/arctic_national_wildlife_refuge/h tml/analysisdiscussion.html

Does anyone know any more about the amount of information we have on ANWR resources and hopw reliable they are?