Something to also keep in mind, is your city and highway driving percentages. The EPA assumes 55% city driving, 45% highway driving for its "combined" MPG rating. If you are getting more or less, then you may not be estimating your city/highway percentages accurately, among other factors of course.
Consumer Reports, newspapers, car websites, and pretty much everyone else has these same findings...gas mileage estimates are extremely random. However, if your Crosstrek should get 26-33 and you're getting 17, that's a problem. Almost every review of that vehicle has the tester getting 27-29. Im in the same situation with my gf's new Elantra GT, albeit not as bad mpg. I'm getting 1 less in the city and 7 less highway. I've even tried cruise, ECO-mode, and normal and found a 2mpg average difference...nothing major. I think people, like me, are realizing that cars should be picked based on everything else and mpg should be estimated at being on the low end of the EPA claim. Before this car we had a 2007 Camry 4 cylinder and the most i got was 25mpg, averaging 20mpg...miserable, but only 2 less than EPA...before then i had a 2003 Accord Coupe that got 30mpg average, so 3-4 above EPA...but 2003 was before the mpg frenzy started as well. Bottom line, just assume the car you buy will average at the EPA's City mileage.
The reason I found this website was my disappointment in MPG of my just purchased 2006 Accord. And I have never trusted EPA ratings, so I wanted real world comparisons, which is what this site sort of provides.
We have a 1995 Accord (daughter's) and a 2001 Camry (mine until I sold it to my brother) -all our vehicles have manual trans. I found I was getting 1.5-3 MPG less than both of the others with the '06. I'll use the 95 Accord as my real bench mark, Accord to Accord. When I complained at a dealer serivce department about this difference I was told the car is heavier and has a bigger engine. Duh! The '06 car weighs 176 lbs more and displacement is up .2 liters on the '95, so I'm not buying that arguement at all. I just put a new cat convt on this week (code P0420) and am watching closely to see any difference.
Just as a reference, my best ever out of the '95 Accord was 33.4 MPG with 167,000 miles on the clock - avg 28 MPG; the 2001 Camry's best was 32.7 MPG with 179,000 on the clock - avg 28 MPG; the 2006 Accord's best so far is 29.6 with 89,000 on it, but at only five recorded fillups I don't trust my avg of 27.1 yet. My daughter says I'm neurotic.
"Driving styles" keep coming up over and over again in this thread. Yet even if people are doing pure city driving, why are they not even reaching the city estimate? The OP is getting 25.0 mpg average in primarily highway driving in a car that is rated at 27mpg for pure city driving. You could assume that they're doing 120 mph in their highway driving, or that they spend most of their time on a drag strip. But a much more realistic guess is since most of the new Imprezas on Fuelly are not meeting the EPA estimates, that maybe the estimates are bogus.
"The EPA figures are only meaningful if your daily commute happens to exactly match the EPA test cycle." That is a misleading statement. The estimates are there as a realistic representation of what most drivers will experience. When the statistics show that certain vehicles don't reach the estimates for a large percentage of drivers, that means the estimates are wrong.
This is becoming all too common with certain vehicles and certain brands. It is out of the realm of "your mileage may vary." The "driving styles" argument is no more credible than Ford Fusion Hybrid MPG ratings when you are talking about conglomerated statistics of dozens or hundreds of cars and drivers.
"The EPA figures are only meaningful if your daily commute happens to exactly match the EPA test cycle."
Well, only if you drive in 25Â°C, no wind, dry roads, no hills(?), with the same rate of acceleration, correct tire pressure and whatnot. Of course with well applied techniques it's possible (not even too hard) to beat official FE figures, but if it's getting harder for newer models then it is a cause for concern.
I easily beat the NEDC estimates on my daily commute. And the NEDC figures are higher than the EPA.
The reason is that I have a particularly good commute for MPG purposes.
Which is the point I was trying to make in my post above.
The same applies here:
"Yet even if people are doing pure city driving, why are they not even reaching the city estimate?" ... but does their city drive match the EPA city drive? Probably not. They are bound to have either more or fewer stops, a longer or shorter distance. Just because they are doing 'city' driving' it does not mean that they will get the same mpg as on the EPA city driving cycle.