I have been trying to fill the tank only when nearly empty, so that I could track mpg vs octane. At 'steady' road speeds, 87 octane was only 1.5 mpg less than 91 octane.
In stop & go driving I got 3 mpg better mileage with 91.
Price differential is about 30% higher for the 91, but my mpg was only about 10% higher.
So I am paying about 20% more for no better results using 91.
With breaking-in a new car and gas prices very low now I've been running 92. In the future I will probably try going with filling the car when about half empty and alternate between 89 and 92 and see how things work out mileage/performance wise.