The Insight's is close but not perfect. Well, at least I know the odometer readings and such are stored in the display unit itself so I assume the mileage numbers are figured there as well considering you can pull the display out of the car, power it up and get the readings that are stored in it.
I would think it's really about 2% off considering the speedo reads 2 mph fast which would result in better mileage readings than you actually get. On the fuel use side of things it measures accurately as far as I know. The problem with trying to determine how exact these things are lies in how you determine mileage. If you do the traditional method you fill your tank, reset the trip odometer, then divide miles by how many gallons you use at the next fill up. When you do this many things come in to play. One of course being how do you know the tank has just as much in it as when you started before? Density of the fuel will even vary by temperature and such. I mean heck, you can fill the Insight up until the pump clicks off, wait a few minutes for the canister to fill then put almost another gallon in.
I've found my comparison of the display to what I figure by dividing miles by gallons to a top off is generally pretty close to what the car displays. I think an issue with the prius is the fact that it uses a bladder to store fuel in rather than a tank. It expands when filled and contracts as it's used. I would think especially when these are new that they would be more difficult to expand all the way causing a fill up to be less than maximum capacity resulting in a higher mileage calculation. I doubt Toyota would release the car with such a large error in it's calculations.
My short answer is this: I have owned 2 Insights, driven many thousands of highway miles, and I measure this stuff meticulously via GPS, fuel receipts, and HP scientific calculator. On both of my cars the odometer read 2.15% fast at 65 mph. However, as a stand alone, the fuel and mpg computer display is extremely accurate; it appears to me to be exactly on the mark. Billy......
I was wondering this today - if mileage is miles travelled divided by fuel consumed, and if miles travelled comes from the odometer (via something that counts each wheel rotation, I assume), then how is the other part of the calculation obtained? Is there a meter in the fuel line that measures fuel flow? Or is the fuel consumption inferred from other measureable elements, such as accelerator depression, RPM, etc.?
[not a hijack]
i've noticed a few times when coming to a stop while auto-stop has engaged: the instant MPG readout shows 150+ until your speedo hits 0mph, then your MPG drops to 0. However, you can still be moving very slowly, and your speed will still show 0mph (another side note, continuing to roll while at "0mph" will cause the car to turn on again - trying to prevent rolling)
so if this above case is true, then the MPG is determined by a digital readout of your speed, and of course fuel consumption. of course, like flipping a coin, there will be an average with a margin of error quite small, it's still interesting to think just how accurate the MPG readout is.