We could have a whole thread dedicated to arguments for and against various protocols for topping off. The arguments against topping off:
* It can lead to the charcoal canister ingesting liquid fuel through the evaporative emissions system, which permanently damages it. Especially in newer vehicles, charcoal canister replacement is an expensive and technically complex operation.
* Newer gas pumps are designed to backswallow fuel that won't fit in the tank. Since the components responsible are downstream of the fuel meter, this leads to the customer paying for fuel that has not in fact been dispensed into his or her tank.
The arguments for topping off:
* Owing to foaming in the tank, backpressure can develop that trips the shutoff before the tank is actually full.
* Owing to unhandy geometry of the filler neck or other limitations in the design of the fuel tank, the shutoff can trip even when the tank is almost empty.
Personally, I don't want to risk damage to my evaporative emissions control system, but on the other hand, I don't want to close a dispensing operation until I am satisfied that the tank really is full, because I also track gas mileage on a per-tank basis and rely on filling to a consistent level to obtain figures I can trust. Although none of my current vehicles has this particular problem, I used to have a 1986 Nissan Maxima that was capable of tripping pump shutoff when just two gallons had been pumped into a nearly-empty tank with nominal capacity of 15.5 gallons.
So, here is my fillup protocol:
* Pump handle locked open (using a hold-open detent) for bulk fill at the slowest available setting until first shutoff.
* Hand feeding of additional fuel (at a variable but generally slow rate) until second shutoff.
* Further hand feeding of fuel until third and final shutoff.
If the feeding time to third shutoff is shorter than the feeding time to second shutoff and the amount of fuel dispensed between first and third shutoff is about 0.25 gallon, I conclude the tank is full and close the dispensing operation. Since I typically pump more than 10 gallons to first shutoff, my maximum potential loss due to fuel reflux is 2.5%. Since fuel tanks are designed with voids equal to 20%-25% of their total volume when full in order to accommodate expansion of the fuel, an extra 0.25 gallons in a tank designed to hold more than 10 gallons (my Saturn holds 12.5 gallons while the Toyota holds 17.5 gallons) will not put me at noticeably greater risk of flooding my evaporative emissions control system. A slowest-as-possible dispensing rate for the bulk fill also minimizes the risk of premature pump shutoff.