I drove over the Santa Cruz mountains today which is mostly uphill for 10.2 miles from about 200 feet elevation to 1800 feet, then mostly down for 12.3 miles back to sea level. I was driving about 60 with no cruise control and temps were 55-60. Here's what I saw:

Bottom of the hill (San Jose side) to the summit, 10.2 miles, 31.0 mpg
Summit to bottom (Santa Cruz side), 12.3 miles, 125.6 mpg
Total mpg: 52.7

Returning home:

Bottom (SC side) to summit: 12.3 miles, 27.7 mpg
Summit to bottom (SJ side): 10.2 miles, 111.5 mpg
Total mpg: 42.0

One slope was slightly longer than the other (10.2 miles vs. 12.3), so it made a big difference on mileage what direction I was going. When the 12.3 mile section was my downhill section I did much better (by over 10 mpg) simply because I was coasting longer.

This is why I think that even when comparing mileage between a bunch of cars of the same make, you can't just assume that the ones with better mileage were driven by more skilled hypermilers, you also have to consider the terrain in which they are driven, which can make a significant difference (assuming all other factors, like temps, speed, etc. are the same).