I don't have the data anymore (it was a 2003 that I sold in 2005, not a 2005). When I owned the mini, I had a reputation for being a bit of a nut about fuel economy. It is my standard practice to record every input of fuel to each of my personal vehicles, along with the date and mileage. I periodically transfer my numbers to a spreadsheet where I plot fuel consumption and mileage vs time. Fuel consumption is plotted using a 3-tank moving average (dotted line) overlaid on the raw data points. Strongest influence generally seems to be the season - weekly avg ambient temperatures from noaa overlay the fuel economy graph quite well if the scales are set appropriately (in the case of my pickup truck, its gradual circle around the drain has been a stronger influence lately). I drive the same route to work every day, at the same times, in very similar traffic conditions. For a long part of the time when I had the mini, there were only 3 stoplights on my 20min drive to work, so I basically cruised at 45mph-55mph the entire way, save 4 periods of hard acceleration. I don't stop for fuel unless I'm worried that I won't make it home from work. In the case of putting 87 fuel in the mini, which asked for 91 fuel, the difference was so striking as to convince me in a single tank that there was a clear difference and that would lose money using 87. I repeated the experiment on several occasions, however, because of regular discussion online about how 87 octane fuel would work just as well as 91, or according to some, better. Most recently (just before i got rid of the Mini) Greg L had convinced me to try again, with the argument that the difference in fuel economy couldn't be greater than the 10% (at the time) difference in fuel prices. Turns out that it was... so I started shopping in earnest for a vehicle with a lower payment and 87-octane fuel. Perhaps the station I filled at had "bad" 87 octane fuel - maybe they used more ethanol in the cheap stuff (would guess the opposite, though) - who knows.