Here's the test schedules that the 2007 EPA test includes:
http://www.fueleconomy.gov/feg/fe_test_schedules.shtml
I do think there are some issues with the EPA's methodology - due to CAFE requirements, the old 1975 test cycles - designed when there was less traffic, and under the 55 mph national speed limit - are still there. Also, stop-start isn't used, meaning some automakers don't want to bring stop-start technology (they fear that Americans aren't ready for stop-start's NVH additions, especially with no gain in EPA ratings.)
I think it'd be best for the EPA to use three sets of test, all the same cycles:
A heavy traffic city surface street test (think NYC)
A moderate traffic city surface street test (able to reach 35 mph consistently, hitting lights about half the time)
A light traffic city surface street test (able to reach 35 mph and stay there)
A 55 mph 2-lane highway cycle
A 75 or 80 mph divided highway cycle
These cycles would be run three times. All three sets would be run at a certain target cabin temperature, with the testing facility refrigerated or heated as required - if it's too hot in the cabin, additional ventilation and/or AC are used. If it's too cold, heat is used. The three sets would be: Average US winter temperature, average US spring and fall temperature, average US summer temperature.
The heavy/moderate/light is necessary, due to multiple definitions of city. For instance, where I live, I have light to moderate city traffic (although I short-trip, killing my fuel economy). But, my results are completely useless to someone living in NYC.