I agree that the accuracy is often questionable, but mine tends to read consistently higher than actual mpg, usually by 1 to 1. 5 mpg. However, the effects of idle, hot-rodding, or coasting down hill tend to mirror the real effects on fuel consumption. I have also concluded that there is some logic in the program as it relates to DTE (distance until empty). The range remaining figure clearly considers the MPG from previous tanks. Mine often shows in the high 500s, even over 600 miles remaining after a fill. When it was new and the mpg was worse, the DTE was much lower. A switch from consistent highway driving to city or vice versa will skew the DTE after a fill up, given a few cycles. A very telling stat can be derived by using the elapsed time function - divide your miles since reset by the hours since reset to derive average speed. If your average is 25 mph, even though you were doing 50 while not at a stop light, the mpg is crappy. Average 50 or better, and you're looking at the best mpg you'll see. All overhaeds should have an average mph, as well as cumulative and tank-to-tank mpg. All the inputs are there, just need a smart chip. Our Lexus LS 430 has that, and it's very useful info. Theasians also found a way to make it scarry-accurate. I kid you not, within a tenth of a mpg or so ... so little it could be attributable to filling inconsistincies at the pump. So I digress ... . bottom line, the overhead on a 3d Gen is very useful if you learn to interpolate what it's telling you. I bet my next pee stop and coasting in to the station on fumes based on it's information. Highly recommended for bypassing a fuel stop in Oregon (no self service, and the cost is therefore too high).