Car computer vs "manual checking" to determine mileage came up in another thread, and I made a somewhat off-topic post there that I'll put here:
This always comes up in gas mileage discussions. In theory, the car's computer system should be the
most accurate calculator of mileage, since it measures the fuel actually being passed through the intake system and miles actually traveled.
Doing it the manual way is fraught with variables that affect accuracy - you would have to somehow drive the same way, return to the same gas pump at the same station, park in the same position for your refill, do it at the same time of day (for gasoline and car temperature reasons), and somehow make sure the pump shuts off at the exact same point in the refueling process. Even then, an air pocket in your tank could throw it off (we used to rock our cars back and forth after filling them and they would sometimes take another half-gallon or more).
My point is that other than a laboratory-level test setup, there is always some degree of inaccuracy in ANY kind of test. Some people believe the car companies purposely build in some error in the car's computer to make the mileage look better, and who knows if that's true.
Some people get all OCD about it, but I've done some testing of my own in the past and I'm comfortable with the mileage shown in the car's computer. You could say it's within my personal margin of error.
