Most all trip computer fuel consumption systems are quite inaccurate, more so on vehicles with smaller (fewer cylinders) engines. This, aside from the imperfect world issues highlighted by
V8eyedoc, is for a number of reasons; a significant one being that most systems have no actual flow rate metering device, such as those used in filling station pumps. Instead the ECU calculates fuel consumption by combining what it knows of:
- the fuel injector's flow capacity which is non-linear across its operating range and further affected by any given device's mechanical tolerances, electro-mechanical performance characteristics, and fuel pressure;
- the operating fuel pressure (often just assumed to be the system design pressure);
- the calculated injector pulse width;
The first two are the least accurate of the lot. And even though the pulse width calculated by the ECU is far more accurate it (when combined with the injector's actual capacity at the actual fuel pressure) makes the final fuel consumption calculation for any single injector at VERY best ±5% accurate for each calculation--more typically ±10%.
In a multi-cylinder engine the error tends to be "averaged out" and generally reduced, and reduced more the more cylinders. Nonetheless the error will remain rather gross making "MPG" displays more of a marketing gimmick than anything to bet money on.
There are aviation and other commercial fuel consumption systems that do use flow rate devices, though these most often report use in gallons or pounds per houir--much more critical information on a ship or in an airplane...