I usually factor a whole tank in my mileage figures, but am happy to play with numbers based on 100 miles as well. If I'm only trying to determine "highway" mileage, and not a combined mileage, I don't mind using the lower numbers.
This way your peak mileage reading is still at least a valid number for your setup. If not completely consistent over time.
I've used as little as 65 miles for testing different things. When I had the time to waste (hard to remember a time...) I would literally go out late at night to avoid the traffic slowing and speeding thing, and run the same route using different setups, with mostly using cruise control.
Different speeds, tire pressures, overdrive vs 1:1, oil viscosities and fuel octane ratings. Just to see if they made any differences.
Sometimes they did, sometimes they didn't.
One interesting note was running my '96 K1500 when it was stock, and after the lift and 35" tires. When stock, it would get it's max mileage (17.5) in overdrive at 60. Shifting to Drive would lose a tiny amount. But after the 35's were installed (same 4.11 gears) my mileage went down to 13 and would not change whether in overdrive or just drive. At least between 60 and 70 anyway.
If I remember it was only about a 400-600 rpm change between the two at 65, but that was a long time ago and I don't really remember.
The biggest gain in fuel economy came after re-tuning it with an early version of the Hypertech OBDII tuner. Never got back to original stock mileage, but at least it got 2mpg back to 15.
Paul