The results are in for my first tank: 27.3 mpg. I take this with a VERY large grain of salt, since the dealer filled it, so I don't know exactly how full they filled it, they probably had no incentive to fill it particularly full, I tend to "top up" more than I should, and the needle seems to be staying on full a whole lot longer on the second tank than it did on the first.
Fuel economy is a complex beast, but as far as I know, the best way to get reliably better or worse gas mileage is to drive a different model car, change driving habits (e.g., speed, acceleration), change driving conditions (e.g., city vs. highway), or change maintenance (e.g., tire pressures). Theoretically the difference in cars that are the same model with the same options should be minimal: the EPA expects repeated, tightly controlled tests on "identical" cars to vary by less than three percent. Driving habits, traffic, and environmental conditions, all of which can vary wildly, probably account for most of the differences in mileage for new cars.
This link from Car and Driver (
http://www.caranddriver.com/features/09 ... es-feature) has an interesting article on EPA estimates that's informative, but not overly technical. It's quite an eye opener. Test conditions are well controlled, but the fox guards the hen house (although the manufacturer has some explaining to do and incurs risk if their figures are too different from an EPA spot check), and the auto manufacturers have a lot of negotiating power on how their cars' features play into testing. I think it's interesting that the vehicles are tested on dynamometers so that wind resistance, which is a MAJOR factor in the real world, does not occur the test, although I'm sure they try to take that into account with all their methodology and complex adjustments. All in all, I find it no surprise that real world results are more variable and almost always lower than the EPA estimates.
The relatively low EPA estimates were already a factor in the "cons" column when I decided on the Versa. If I had known that people where having a greater than average decrease in "real world" mileage vs. the sticker it may have discouraged me a little more. I can only wonder how much the mileage will decline as the car ages--there was a noticeable decline in gas mileage within the first 100,000 miles for my other cars. On the bright side, it looks like my Versa might do quite well with real world vs. sticker mileage compared with the other cars I've owned. The second fill up should give a far better feel as to whether this will be the case.