To keep your car in the best condition, it’s important to drive it regularly. Those in rural areas or who have long commutes to work spend a lot of time inside their cars. Drivers in these situations typically want a car that’s good on gas. You can find any model’s fuel economy ratings on the automaker’s website.

However, some drivers have noticed that these numbers are not always accurate. Discrepancies between fuel economy ratings and the car’s actual mileage can even lead to lawsuits in some cases. How do automakers determine these numbers, and how do we know which ones are correct?

How large are these inaccuracies? 

As this Reddit thread proves, cars getting different fuel economy numbers than advertised is not uncommon. This isn’t always a bad thing. One user reported that their car gets 35 mpg on the highway instead of the estimated 31 mpg. Another user’s Jaguar F-Pace is rated at 26 mpg on city roads, but they regularly average 40 mpg.

For some users, their real-world driving numbers are far lower than advertised. One user reports getting only 11 mpg in the city on a vehicle that’s rated for 19 city mpg. Why are some drivers able to get better scores than others?

How is gas mileage calculated?

Contrary to popular belief, the EPA doesn’t test every new vehicle for fuel-efficiency. Only 10 percent of vehicles end up being tested, and most of these are selected randomly. For the vast majority of cars, the EPA relies on the automaker’s provided estimates. Each manufacturer tests its own vehicles.

The EPA uses a simulator called the dynamometer to get the most accurate fuel economy numbers. The “city” setting operates under the assumption that the engine is started cold and will experience frequent idling in congested traffic. The “highway” setting recreates a scenario where the engine is already warmed up and making no stops.

Are real-world driving numbers more accurate?

Websites like Edmunds and Consumer Reports include real-world numbers with each car they review. Given how the EPA’s test is conducted, it’s easy to see why the numbers could be skewed in certain conditions. This is especially true for city driving since not everyone has to deal with heavy traffic on a daily basis.

Differences in the numbers can also be attributed to driving style. If you’re an aggressive driver, your car’s fuel economy will be lower than average. A car’s fuel economy can also be lower due to the weather, low tire pressure, or towing large trailers. Consumer Reports found that the numbers can also change depending on the car’s engine.

What about hybrids and EVs?

Cars that have some sort of electric powertrain are naturally better at saving gas. Many hybrids use electric power only at low speeds, which would improve city driving numbers drastically. You can find the automatic start-stop system on cars like the Honda Civic Hybrid and the Toyota Prius.

The EPA has a specialized test for hybrids, but it’s very new and still prone to error. Since hybrids are even more sensitive to driving behaviors, you may frequently see cases where real-world fuel economy is drastically different on some hybrid cars.

Can we trust the manufacturers’ numbers?

A good percent of drivers still manage to reach the numbers reported by the automakers. In general, we can assume that these figures aren’t too far off the mark. If you want to be cautious, always assume that your car will reach the low end of its predicted estimates.

There’s a reason the phrase “your mileage may vary” exists. Don’t buy a car solely based on its fuel economy, especially if you’re one of those drivers with a lead foot.