With the exception of the Corvette that Mr. Slant mentioned (Which I think was more of an emission issue then a fuel mileage one) I do not know of any car that has a computer that does this.
It really does not make sense if you think about it. With emission standards being what they are now, they don’t inject any extra fuel anytime if they don’t have too. The car always runs as lean as possible.
“YMMV” is not just an abreviation used on internet message boards. 
Here’s a previous post about how the EPA rating is derived.
FWIW in my big truck, I get EPA mileage in the city easily (12 MPG). On the highway, I can’t meet EPA; the thing’s 7.5 feet tall and at 80mph there’s just no way to do it. I did an experiment several months back (and posted it here) whereby at exactly 100k/h, I got 17mpg (yeah, mixed units), which is 1mpg higher than the EPA estimate. However It’s impossible to less than 80mph, because the only thing worse than a jackass, aggressive, speeding, gigantic SUV is one that drives 20mph slower than everybody else. So at “normal” commuting speeds (virtually all freeway) I typically get 14 or so, but that’s just an estimate, because I don’t drive it often enough to get a good measurement (duh, that would be wasteful just for the fun of it).
While it was done to increase fuel economy, in the real world it worked just fine. The engine had so much torque that you really didn’t need to go through all those gears to accelerate. In normal driving (not stop light Grand Prix) you wouldn’t notice.
Well, it was to my understanding that car engines could make a tradeoff between fuel efficiency, responsivness, power and a whole bunch of other factors. Under normal driving conditions, people would prefer a more responsive car at the expense of fuel efficiency. But in test situations, you can eke out a few more MPG by completely sacrificing every other factor.
If my car isn’t “responsive” enough at 25% throttle, I can fix that with 50% throttle.
If automakers don’t meet government-mandated Corporate Average Fuel Economy goals, they pay through the nose. As a result, every mass-market automaker selling in the US tunes the bulk of their models for fuel economy first.
Car manufacturer’s hands are tied on this one. They are not allowed to quote any fuel economy figure other than the EPA’s rating. “YMMV” is as far as they can go in telling you what you might really see.
As for gaming the tests at least one maker was found to have engine control software that recognized the EPA test sequence and modified the engine parameters to improve emissions. I’ve not heard of it provably being done for MPG numbers.
And if we did this every customer that hit that flat spot would come shove that car stright back up our ass. 
I also think you are way over stating the amount that might be gained. Back in the 60’s maybe you could have done this. Cars that ran the Mobil ecomony run got unreal gas mileages when compared to what you and I would get with the same car. This was due to these cars being driven by pros, and having special optional equipment to improve gas mileage like a real loooooooonnnnnnnnngggggg rear end ratio so the car is just off idle at freeway speed.
But nowadays with the emissions requirements being what they are, cars don’t have that kind of room to move.
The only time I have ever heard this claim has been on the internet. I have never heard this claim in the industry, including many many talks with engineers that work on fuel systems.
I’ll try and give the head of our factory engineering group a call today and ask him, I’m sure he is up for a laugh.
But if he laughs about it, I am going to need one hell of a good cite to get me to buy this. I think this is same relm as the 200 MPG car
No, the computer would switch to fuel efficient mode ONLY when it detected the EPA test being run. Once the run is finished, it would switch back to the normal, maginally less efficient mode.
Again, the cars supplied to the EPA tests are not necessarily the same as the production models. They have to supply prototypes in many cases due to the need of having the tests done in advance of full blown production. In particular, the control software on modern computers can change during the model year to fix/enchance things. Cars being shipped with different control software from the ones in the EPA test should be expected.
This is one of CR’s standard rants about what is wrong with the EPA tests.
The EPA sticker is just a piece of “car spam” you have to scrape off the window after you buy the car.
So I will ask you the same question I often ask my students.
How do it know? 
How does the control unit know that an EPA mileage cycle is being driven?
The legend I’ve heard is that the computer recognizes the first three “moves” or whatever of the test and switches over. Presumably it’s a few hour long test and once it sees the set pattern:
Acelerate at 3.3 mph/second up to 32 miles per hour, come to complete stop, accelerate to 40 miles per hour, stop, accelerate to 30, etc., then it would switch over.
I don’t know, I doubt it’s real.
It’s not a few hours long.
Also from your location (NYC Metro Area), you can’t get gasoline, but a blend of gasoline and some other ‘fuel’ which also contains oxygen, which you normally get from the air - Yes along with fuel, you have to carry around a part of the ‘air’ in your gas tank. This means that your gallon of fuel does not have as much energy in it as a gallon of gasoline, which would effect milliage.
Cars and trucks today are making an astounding amount of horsepower while still getting good EPA fuel estimates. They do this through computer technology. When you aren’t accelerating hard, the onboard computer can do everything from retarding the timing, leaning the mixture, or even in some vehicles shutting off half the cylinders.
So, you can find a Corvette with 500 HP that gets over 20mpg. But guess what? Drive it hard, and alll that gas-miserly technology goes out the window, because it simply takes a lot of energy to produce 500 horsepower at the flywheel. So cars today are more sensitive to driving technique than they used to be.
And there is a huge difference in gas mileage with even a slight increase in highway speed. I just saw an episode of Motorweek, where they sent two drivers out in identical vehicles (Chrysler 300C’s). One drove ‘agressively’, while the other drove normally. The biggest differerence came in highway driving. The more ‘aggressive’ driver drove something like 10 mph faster than the other guy, who drove the speed limit, and it cost him 25% in fuel economy.
Is that the case in eastern Long Island? I typically don’t buy fuel in NYC.
“Metro Area” in public policy/government use means the extended vicinity of the city, even including neighboring suburbs, sometimes in the next state.
Here’s a national map of various “local gas” varieties:
**PDF WARN
http://www.econbrowser.com/archives/2005/08/boutique2.gif
**PDF WARN
Here’s an EPA chart on the matter:
**PDF WARN
http://www.epa.gov/otaq/regs/fuels/420b05012.pdf
***PDF WARN
The way I read that chart, all of NY state gets the same gasoline spec.
OK, I finally caught up with my buddy. I asked him about how the EPA numbers are arrived at. In so many words the carmaker supplies coast down numbers to the government. An instrumented vehicle is driven to 60 mph and allowed to coast down to 10 mph (or there abouts) this supplies friction and wind resistance numbers. The track this is done on is perfectly flat. I have seen the one at our proving grounds in Arizona and I was told that the surface is level within a 1/4" over a 2-mile stretch.
The EPA as part of the emissions testing looks at the fuel consumption. That consumption + the coast down numbers provides the gas mileage numbers. The combined number is based on a combination of the two and a mathematical formula, which he did not recall off the top of his head.
Then I asked him about cheating the test. He laughed for a good 30 seconds. He then said that doing that would be referred to as a defeat device and is strictly illegal. If the government discovered such a device the fines and penalties are huge. :eek:
He stated that there are absolutely no defeat devices on our cars. Of course he can’t speak for other carmakers, but the penalties for getting found out are pretty severe and would tend to convince carmakers from going there.
SciAm April 2006 discusses this very issue. The article indicates there is potentially a large disparity between testing conditions and actual conditions, especially with the hybrids where the type of driving has a direct impact on how the hybrid engine operates.