Why is my MPG lower than the official figure?

Everybody knows the brochure figure is optimistic. Almost nobody knows by how much, because the number that would tell you is the one thing car makers don't publish. We've got it: 8,216 engines with an owner-reported average sitting next to the official laboratory figure, and the gap is both bigger and more predictable than you'd think.

Why is my MPG lower than the official figure?

Because the official figure is measured in a laboratory, on a fixed test cycle, with none of the things that actually burn your fuel. That's not a scandal, it's the point: every car is measured identically so the numbers can be compared. It just makes them a poor guide to what you'll spend.

The size of the miss depends almost entirely on which test your car was measured on. Across our Real MPG database, the typical car tested on the old NEDC cycle returns 82% of its official figure. The typical car tested on the newer WLTP cycle returns 94%. So an NEDC-era car missing its brochure number by roughly a fifth is completely normal, and a WLTP-era car doing the same is not.

How big is the gap in practice?

Here's the whole distribution, taken from every engine in the database that has both an owner average and an official figure. NEDC and WLTP are kept apart because mixing them is meaningless.

Two things jump out. Two thirds of NEDC-era engines land between 70% and 89% of the brochure figure, which is an 11 to 30% shortfall and is the normal case, not a fault. And about one engine in ten across the whole database actually beats its official number, which is worth remembering before you assume your car is broken.

What's the difference between NEDC and WLTP?

They're two different laboratory test cycles, and WLTP is the harder one. According to the Vehicle Certification Agency, WLTP replaced NEDC for official fuel consumption and CO2 figures in September 2017 and became mandatory for all new petrol and diesel cars by September 2018. The VCA's own description is that WLTP gives a closer representation of real-world figures than NEDC did, while being clear that both are still laboratory tests.

Our data says the VCA is right, and puts a number on it. Same database, same owners, same roads: 82% against NEDC, 94% against WLTP. That 12 point difference is the size of the improvement.

What this means when you're shopping:

• A 2016 car's official figure and a 2020 car's official figure are not comparable. The older one was measured on the easier test, so it looks better on paper than it deserves.
• Cars sold across the changeover often have both figures quoted, which is why the same model can appear with wildly different mpg claims depending on where you read it.
• A percentage of official only means something next to its own test. On our tables, an asterisk marks a figure measured against WLTP. 95% with an asterisk and 95% without are not telling you the same thing about the car.

Do small engines miss by more than big ones?

Yes, and this is the finding that surprised us. Taking every NEDC-tested engine where we can read the capacity off the variant name, 4,841 of them:

The middle of that table is fairly flat, so this isn't a clean line. But the two ends are clear: the smallest engines miss their official figure by the most, and engines over 2.0 litres come closest to theirs. That's the downsizing story in one column. A tiny turbocharged engine flatters a gentle test cycle because it barely has to work, then spends its real life boosting to keep a 1,300kg car moving, which is when it drinks.

The Ford Fiesta 1.0T EcoBoost 100 from 2013 to 2017 is the textbook case. Official 65.7 mpg, owner average 46.3, which is 70% and a shortfall of about 19 mpg. The same generation's 1.25 petrol, an older and simpler engine with a lower official claim of 54.3, returns 41.1 in the real world, which is 76%. The little turbo has the better brochure number and the worse honesty rating.

Do the same sum on the WLTP-tested engines and the size effect very nearly disappears, with every band landing between 91% and 96%. The harder test is much harder to flatter with a small engine, which is exactly what it was designed to fix. It also reshuffles the economy rankings completely, which is why the most economical used cars in the real world are not the ones with the best brochure figures.

What actually costs you the fuel?

The test cycle can't include any of this, and between them they explain most of a normal gap:

• Cold starts. An engine is at its least efficient before it reaches operating temperature, and it runs a richer mixture to get there. A five mile commute can be most of a cold start.
• Short trips generally. Nothing gets warm: not the engine, not the oil, not the catalytic converter, not the diesel particulate filter.
• Speed. Drag rises with the square of speed, so 80 mph costs a lot more than 70, which costs more than 60. This is the single biggest lever most drivers have.
• Load and shape. A full boot, four passengers, a roof box, a towbar, a bike rack. A roof box left on all summer is money out of the tank on every journey.
• Tyres. Underinflation raises rolling resistance. Fitting a wider or grippier tyre than standard costs economy permanently.
• Cold weather. Denser air, thicker oil, longer warm-ups, heated everything, lights on. Winter figures are always worse than summer ones, on any car.
• Right foot. Acceleration and braking are the same energy converted into heat. The more of it you do, the more fuel you buy.

When does a bad MPG figure mean something's wrong?

When your car is well below what other owners of the same engine report, rather than below the brochure. That's the comparison that matters, and it's the one the official figure can't give you.

If you're 20% under the official number on an NEDC-era car, you're normal. If you're 20% under what other owners of your exact engine average, something is usually going on, and a handful of faults show up as fuel economy before they show up as anything else: a diesel particulate filter stuck in a regeneration loop, a coolant thermostat stuck open so the engine never warms properly, tired injectors, a dragging brake caliper, a boost leak on a turbo engine. Several of those are documented on our common problems pages for the models they're known on, including Ford Focus known faults and VW Golf known faults.

That's also the practical reason to check a car's real-world figure before you buy rather than after. If the model's owners report 45 mpg and the seller mentions the car does "about 33", that's either a hard-driven car, a short-journey car or a car with a problem, and all three are worth knowing about before you hand money over.

Can a check tell me what a used car will cost to run?

Some of it, and here's the part it can't: nobody can predict your fuel economy, us included. It depends on your commute, your right foot and your local roads more than on anything in a database, which is why the crowd figures above are the honest source for that half. On provenance records we're in the same position as everybody else too. Finance comes from the lenders, theft markers from the police, write-offs from the insurance industry, and every check in the country reads the same registers.

What we add is the reading of the advert itself, which no plate lookup has ever seen, along with the full DVSA MOT advisory history and a database of known faults matched to the exact model, engine and year. The Smart Check turns that into the running-costs picture for a specific car: estimated road tax, insurance group and typical service costs, a 12-month maintenance forecast, and the model's known faults priced up. That last figure is the one that dwarfs fuel. Saving £150 a year at the pump doesn't survive one dual mass flywheel, and the report tells you which of those the car in front of you is due.

It won't spot unrecorded accident damage or judge the quality of a past repair, because those aren't in any database at any price. And it won't tell you your mpg.

So the sensible order is: look up the model's real-world figure on Real MPG to see whether it suits your mileage, then run the free scan on the actual advert to see what the car itself is carrying. The scan needs no card and no password, and if it turns up something worth going deeper on, the free report upgrades on the same link afterwards without re-running anything.

The short version

Official figures come from a laboratory, so they're for comparing cars rather than predicting your spend. On the old NEDC test the typical car returns 82% of its official figure. On WLTP, which replaced it from September 2017, the typical car returns 94%. Small engines miss by the most on NEDC and barely at all on WLTP. And the number worth measuring yourself against isn't the brochure, it's what other owners of your exact engine actually get.