Everyone knows the consumption figure in a car brochure is optimistic. What almost nobody knows is that since 2021 the car itself has been keeping score, that the European Commission has been collecting those recordings from hundreds of thousands of vehicles, and that it has published the size of the gap. The number is larger than the folklore assumed, and for one category of vehicle it is not a gap at all: it is a different order of magnitude.
The device you did not know you had
It is called OBFCM: on-board fuel and energy consumption monitoring. It records what the vehicle actually consumes over its life, not what a laboratory measured once.
It has been mandatory since 2021 for all new combustion and hybrid cars, plug-in hybrids and most vans, with heavy vans following from 2022. The legal basis is Commission Implementing Regulation (EU) 2021/392, implementing the CO2 standards in Regulation (EU) 2019/631.
This is a genuinely unusual piece of regulation. Rather than arguing about whether the official test represents reality, it requires the fleet to measure itself and report back.
What the fleet reported
The Commission published the first analysis on 18 March 2024, as COM(2024) 122 final. The dataset covers vehicles first registered in 2021 with data collected in 2022: 617,194 cars and 6,667 vans.
For conventional cars the divergence between real-world and the official WLTP figure was:
Petrol cars: 23.7% higher in the real world, 34.6 g CO2/km above the type-approval value.
Diesel cars: 18.1% higher, 27.8 g CO2/km above.
Then the plug-in hybrids, where the word "gap" stops being adequate:
Petrol plug-in hybrids: 238%, about 100 g CO2/km above the official figure. Diesel plug-in hybrids: 312%.
Worth stating fairly: the Commission notes this is roughly half the divergence seen under the older NEDC procedure. WLTP genuinely improved things. It did not close the gap.
Why plug-in hybrids are so extreme: the official figure assumes a particular balance of electric and engine driving. A plug-in hybrid that is rarely plugged in is simply a heavy petrol car carrying a battery. The test result is not fraudulent: it describes a usage pattern that a great many owners do not follow. This is a case where the official number can only be right if the driver behaves in a specific way, and the fleet data shows most do not.
Two credible studies, two different answers
There is a second body of evidence, and it does not agree with the first. Reporting only the larger number would be easy and misleading, so here is both.
The ICCT analysed roughly 162,600 combustion and hybrid cars using data from the spritmonitor.de logging platform merged with European Environment Agency type-approval records, covering registration years 2011–2022. It found the WLTP gap growing from 7.7% in 2018 to 14.1% in 2022, an increase of more than 80% in the five years since WLTP became mandatory in September 2018. For 2021 it reported roughly 13% for petrol and 11% for diesel.
The Commission's figures for that same 2021 registration year were 23.7% and 18.1%. That is a wide disagreement between two credible sources.
The methods differ, and the difference points one way. Spritmonitor data is self-reported by drivers who choose to log every tank, a group unlikely to be representative, and one that tends to be more conscious of consumption than average. OBFCM is measured by the vehicle itself across whatever the fleet actually does. On the question "what does a typical car really use", the self-selected sample is the weaker instrument.
We are not going to pretend that resolves it completely. What can be said is that both find a substantial gap, both find it in the same direction, and the more representative dataset finds the larger one.
What to do with this
- Add roughly a quarter to the official petrol figure when estimating running cost, and about a fifth for diesel. That is the fleet-measured average, not pessimism.
- Treat a plug-in hybrid's official number as conditional. It describes a car that gets plugged in regularly. If you cannot charge at home or at work, the honest planning figure is much closer to a petrol car of the same weight.
- Compare like with like, and check the procedure. A pre-2018 NEDC figure and a WLTP figure are not comparable. NEDC numbers are optimistic by roughly twice as much.
- Your own history beats every published figure. Litres divided by kilometres across several tanks is a measurement of your car, your routes and your driving. Nothing in a brochure competes with that.
- Do not read the gap as scandal by default. A laboratory test has to be reproducible, which means it cannot include your hills, your traffic, your roof box or your winter. Some divergence is structural. The size of it is the fair question.
Why this is the same argument we keep making
We write a lot about the difference between a published figure and an observed one, and this is the cleanest example of it anywhere in motoring.
The WLTP number is not a lie. It is a careful, reproducible measurement of a specific procedure, and it is genuinely useful for comparing two cars against each other. It simply is not a measurement of your car on your commute, and for years the only way to know the difference was anecdote.
Then Europe required the fleet to measure itself, and the anecdote turned into 617,194 data points. The answer was not that the official figure was roughly right, and not that it was fabricated. It was that reality ran about a quarter worse for petrol, and several times worse for one category that had been marketed hardest on its efficiency.
That is the whole case for observation over estimate, made by a regulator rather than by us. It is also why every price and status on our map carries the minute a real person reported it, and why we show nothing rather than a plausible-looking figure when nobody has. An estimate that has never met the road is a different kind of fact from a measurement, and the honest thing is to keep them visibly apart.