Electric, petrol, diesel or hydrogen: which pollutes more?

In short
Count the whole life cycle, not just the exhaust. Then an electric car in the EU emits 63 to 70 grams of CO2 equivalent per kilometre, and a petrol or diesel car about 235 grams. The battery makes manufacturing more carbon-intensive. In the EU, an electric car makes that up after about 17,000 to 22,000 km. At 12,000 km a year, that is one and a half to two years. In practice, a plug-in hybrid emits far more than it does on paper. A hydrogen car only scores well with green hydrogen. That takes a lot of electricity. Every figure depends on the electricity mix, the lifespan and real-world consumption. The assumptions are given next to each figure.

Common claims: what is true?

What counts in a life cycle?

A life cycle analysis counts emissions from construction to scrapping. It has four blocks:

• Building the car, with its battery or hydrogen tank, and processing it at the end of its life.
• Producing and delivering the fuel or electricity. Specialists call this "well-to-tank", from the source to the tank. For petrol, that means drilling, refining and transport. For an electric car, it means the power station and the grid.
• Driving. This is called "tank-to-wheel", from the tank to the wheel. Only this block comes out of the exhaust.
• Maintenance, such as tyres, oil and spare parts.

Where you draw the line decides who comes out well. Counting only the exhaust gives an electric car zero grams: too good to be true. Counting only manufacturing makes it worse than a petrol car: too harsh. Only all the blocks together give an honest picture.

The figures below come mainly from the July 2025 update of the International Council on Clean Transportation (ICCT), a non-profit research organisation. The European Climate Foundation supported this report. Researchers at the Joint Research Centre (JRC) of the European Commission reviewed an earlier version. The report calculates for an average mid-size car from 2025 in the EU. The unit is grams of CO2 equivalent per kilometre (g CO2e/km). It also includes other greenhouse gases, such as methane, converted into CO2.

Four choices shift the result strongly:

• Lifespan. The ICCT assumes 20 years and 240,000 km (12,000 km a year). With 15 years and 180,000 km, an electric car's emissions rise by 22% (14 grams per km). For petrol and diesel cars, they rise by 4 to 5%.
• Real-world consumption. A petrol car uses 6.9 litres per 100 km in practice, against 5.7 litres in the official test. An electric car uses 20.2 kWh per 100 km, against 16.2 kWh in the test.
• Electricity mix (the sources the electricity comes from). A car from 2025 drives until 2044, while the grid gets cleaner. If you use the 2023 mix for 20 years, the electric car's emissions come out 58% (37 grams) too high.
• Battery. The ICCT assumes 53.4 kWh and 72.8 kg CO2e per kWh. That is 3.9 tonnes for the battery. The figure is the weighted average of factories in China (56%), Europe (30%), South Korea, Japan and the US. The ICCT deducts nothing for recycling.

If you choose every assumption in the least representative way, you put combustion-engine cars 7 to 11% too low, plug-in hybrids 32% too low and electric cars up to 64% too high. That may explain why some studies reach a different conclusion.

The result in figures

Life cycle, mid-size car (EU, 2025), 240,000 km, real-world consumption:

Source: ICCT, July 2025. The 63 comes from the JRC scenario for the electricity mix, the 70 from the Stated Policies scenario of the International Energy Agency (IEA). The ICCT does not calculate blue hydrogen (natural gas with CO2 capture). E-fuels are fuels made with electricity. According to the ICCT, a combustion engine running on e-fuels can in theory emit as little as an electric car. The supply for road transport is uncertain, though, and the price is high.

Look at the share of construction for the electric car and the share of the exhaust for the combustion cars. The ICCT publishes the split by phase only as a bar chart, so we derived it from its input data.

Emissions over the whole life, per kilometre

Grams of CO₂e per km for a medium-sized car in the EU, split by phase of the car's life.

The split by phase is derived from ICCT tables and is accurate to about 1 g/km. Only the totals appear like this in the ICCT report.

• Building the car
• Building the battery
• Hydrogen tank and fuel cell
• Maintenance
• Making and moving fuel or electricity
• Tailpipe

After how many km is an electric car better than a petrol car?

According to the ICCT: about 17,000 km (EU grid). Our own recalculation: 17,275 km. The Belgian grid is calculated at 150 to 210 g CO₂e per kWh (Elia). The comparison is a petrol car of the same class.

Sources

• ICCT (Negri, Bieker), Life-cycle greenhouse gas emissions from passenger cars in the European Union: a 2025 update and key factors to consider (opens in a new tab) (2025 ) advocacy group
• Elia Open Data, Production-based and consumption-based CO2 intensity Belgium (ods192) (opens in a new tab) (2026 ) primary source

Our own diagram, based on the ICCT (July 2025).

Show the figures as a table

Emissions per phase and in total, in g CO₂e per km

Powertrain	Building the car	Building the battery	Hydrogen tank and fuel cell	Maintenance	Making and moving fuel or electricity	Tailpipe	Total
Petrol	30	0	0	5	48.8	151.2	235
Diesel	30	0	0	7	53.4	143.6	234
Natural gas	31.7	0	0	5	36.3	130	203
Hybrid	30	0	0	5	37.3	115.7	188
Plug-in hybrid	32.9	4.2	0	5	36.5	84.4	163
Electric (EU electricity mix)	27.1	16.3	0	4	15.6	0	63
Electric (EU mix, IEA scenario)	27.1	16.3	0	4	22.6	0	70
Electric (renewable electricity only)	27.1	16.3	0	4	4.6	0	52
Hydrogen car (green hydrogen)	27.1	0	7.5	4	11.4	0	50
Hydrogen car (hydrogen from natural gas)	27.1	0	7.5	4	136.4	0	175

How the break-even point is calculated

Input	Value
Extra emissions from building the electric car (tonnes CO₂e)	3.2
Petrol car: fuel chain without building (g/km)	200
Maintenance, petrol and electric (g/km)	5 / 4
Electric car consumption (kWh/km)	0.202
Electricity, EU average (g CO₂e/kWh)	78
Break-even according to the ICCT (km)	17,000
Break-even, our own recalculation with EU electricity (km)	17,275
Electricity, Belgium (g CO₂e/kWh)	150 to 210
Break-even with Belgian electricity (km)	18,746 to 20,179

In the Global EV Outlook 2026, the IEA calculates with figures for the whole world. The advantage is smaller than in the ICCT's figures, but the direction is the same. A mid-size electric car emits on average more than 55% less than a petrol car, over 15 years and 35 km a day (over 190,000 km), using the Stated Policies scenario for electricity.

And in 2030?

In this update, the ICCT does not publish a separate figure for a 2030 car. The following is known:

• According to the ICCT, the European electricity mix (including the construction of power stations) falls from 264 grams per kWh (g/kWh) in 2022 to 88 to 113 g/kWh in 2030. Those two values come from IEA scenarios, not from measurements.
• Batteries from European factories are cleaner. For an NMC battery (nickel-manganese-cobalt), the ICCT assumes 60 kg CO2e per kWh in Europe and 80 in China. In 2024, 30% of the cells for European cars came from Europe.
• Combustion-engine cars are getting only slightly cleaner. Their real-world emissions fell by 1% on average between 2018 and 2023. The official emissions of all new cars fell by 28% and the real-world emissions by 15%. That fall comes from electric cars (ICCT, June 2026).

The CO2 emissions per kWh of Belgian electricity, which Elia (the Belgian high-voltage grid operator) publishes, do not fall in a straight line. The consumption average (including imports) was between about 70 and 150 grams per kWh from April to October 2025. In the same months of 2026, it was between about 150 and 210 grams. This article does not look into the cause. Elia does not say whether the construction of power stations is counted. The ICCT does count that construction. Elia's figures may therefore be lower than the ICCT's.

From how many kilometres does the electric car win?

At construction, an electric car with its battery emits 10.4 tonnes of CO2e in total, against 7.2 tonnes for a petrol car. The ICCT counts 6.5 tonnes for the car and 3.9 tonnes for the battery. The difference is 3.2 tonnes. An electric car makes that up by emitting less on the road.

Worked example (illustrative, with figures from the ICCT report)
A petrol car emits about 200 grams of CO2e per km for its fuel (production and combustion). That is 235 minus 35 grams for construction and maintenance: 7.2 tonnes divided by 240,000 km is 30 grams, plus 5 grams for maintenance. An electric car uses 0.202 kWh per km. At 78 g/kWh (the ICCT average for 2025 to 2044), that is 0.202 × 78 ≈ 16 grams per km. Saving: 200 minus 16 is 184 grams per km. The extra 3.2 tonnes is 3,200,000 grams. Divided by 184, that gives about 17,400 km. The ICCT itself gives about 17,000 km. This sum leaves out the small difference in maintenance (5 against 4 grams).

The same sum for other grids, each time with a fixed emission per kWh. All rows assume 200 grams for petrol, 0.202 kWh per km and 3.2 tonnes extra:

With electricity at 78 to 260 g/kWh, the extra emissions from construction are made up after about 17,000 to 22,000 km, or after one and a half to two years at 12,000 km a year. Elia's Belgian figures fall within that range: at 150 to 210 g/kWh, it is about 19,000 to 20,000 km. Critics have a point on a grid with coal only (1,023 g/kWh): there, an electric car gains nothing in this sum. Real grids are lower. The IEA writes that electric cars also emit less over their lives than comparable petrol cars in India, China and South Africa.

Keep your old car or replace it?

Your old car has already emitted its manufacturing emissions. A new electric car starts with 10.4 tonnes of CO2e. With Belgian electricity of 2026 (171 grams of CO2 per kWh, Elia), it emits about 38.5 grams per km, and your old car 216 to 242 grams for the petrol and diesel cars, 158 grams for the recent hybrid.

The break-even point per type of car, with the years at 5,000, 14,172 and 20,000 km a year in brackets (14,172 km is the Belgian average, Car-Pass 2025):

• Petrol car from around 2005: 51,000 km (10.2 / 3.6 / 2.6 years)
• Diesel car from around 2012: 56,500 km (11.3 / 4.0 / 2.8 years)
• Petrol car from around 2018: 58,500 km (11.7 / 4.1 / 2.9 years)
• Economical recent hybrid: 87,000 km (17.4 / 6.1 / 4.4 years)

For people who hardly drive (5,000 km a year: 10 to 12 years) or own an economical recent hybrid (87,000 km, about 6 years at the Belgian average), the new electric car needs much longer to repay its manufacturing emissions. Whether keeping your car is then greener depends on how long it will still run.

Three limits. The rebound effect (people who drive electric sometimes drive more) is not in the figures, and we found no Belgian source. How long your old car will last is unknown. For a used electric car, we found no figure for its manufacturing emissions. Those are already emitted, but that is reasoning, not a number.

The Carbon Brief fact check of 24 October 2023 asks the same question, but its chart applies to the UK, not to Belgium.

The chart shows cumulative emissions from today. Look at where the line of your old car rises above that of the new electric car.

Keep your old car or buy a new electric one?

The new electric car starts with the emissions of its construction; your old car starts at zero but emits far more per kilometre.

Open circle: the break-even point, where the new car catches up with the old one

Production: 10.4 t. Calculated for a new medium-sized electric car and electricity at 171 g CO₂e per kWh (Elia, 2026). If the electricity is cleaner or dirtier, the break-even point moves.

Our own calculation on published data. Carbon Brief (UK) is the model for the approach, not the source of the numbers.

The Elia figures are consumption-based and cover only part of 2026. Elia does not say whether building the power plants is counted. On an ICCT (life-cycle) basis the value may be higher.

What the calculation leaves out

• The calculation counts the production of the new car in full as extra emissions. If you would buy a car in a few years anyway, that production only moves in time and is not extra. This makes the calculation favourable to the claim that you are better off keeping your car.
• Rebound: people who drive electric sometimes drive more. That effect is not in the figures. We found no source that measures it for Belgium.

Years to break even, depending on how far you drive each year

Old car	Break-even (km)	5,000 km a year	14,172 km a year	20,000 km a year
Petrol, 2005	51,000	10.2	3.6	2.6
Diesel, 2012	56,500	11.3	4	2.8
Petrol, 2018	58,500	11.7	4.1	2.9
Hybrid, recent	87,000	17.4	6.1	4.4

Sources

• ICCT, Life-cycle greenhouse gas emissions from passenger cars in the European Union: a 2025 update (opens in a new tab) (2025 ) advocacy group
• Elia Open Data, Production-based and consumption-based CO2 intensity Belgium (ods192) (opens in a new tab) (2026 ) primary source
• Car-Pass, Jaarverslag 2025 (opens in a new tab) (2026 ) primary source

Our own diagram, based on the ICCT (2025), Elia (2026) and Car-Pass (2025); the calculation is ours.

Show the figures as a table

Break-even and years, all annual mileages

Old car	Break-even (km)	Range over the electricity scenarios (km)	5,000 km a year	7,903 km a year	10,975 km a year	14,172 km a year	20,000 km a year
Petrol, 2005	51,000	46,700 to 53,100	10.2	6.5	4.6	3.6	2.6
Diesel, 2012	56,500	51,300 to 59,000	11.3	7.1	5.1	4	2.8
Petrol, 2018	58,500	52,900 to 61,200	11.7	7.4	5.3	4.1	2.9
Hybrid, recent	87,000	75,300 to 93,200	17.4	11	7.9	6.1	4.4

Total emissions (tonnes CO₂e) after a number of km

Distance (km)	New electric car	Petrol, 2005	Diesel, 2012	Petrol, 2018	Hybrid, recent
0	10.4	0	0	0	0
10,000	10.8	2.4	2.2	2.2	1.6
20,000	11.2	4.9	4.5	4.3	3.2
30,000	11.6	7.3	6.7	6.5	4.7
40,000	11.9	9.7	8.9	8.7	6.3
50,000	12.3	12.1	11.1	10.8	7.9
60,000	12.7	14.5	13.4	13	9.5
70,000	13.1	17	15.6	15.1	11.1
80,000	13.5	19.4	17.8	17.3	12.6
90,000	13.9	21.8	20	19.5	14.2
100,000	14.3	24.2	22.3	21.6	15.8

Plug-in hybrids in practice

A plug-in hybrid (PHEV) has a petrol engine and a battery that you charge from a socket. The official test assumes a lot of electric driving. For a PHEV with 60 km of electric range, the test assumes 80% electric driving. In practice, it is about 45 to 49% for private cars and 11 to 15% for company cars. The ICCT and Germany's Fraunhofer ISI found this in 2022 among about 9,000 vehicles.

• The ICCT's test value is 1.1 litres of petrol and 15.2 kWh per 100 km. The real-world average is 3.8 litres and 11.9 kWh.
• Since 2021, the European Commission has measured real-world consumption with a meter in the car (on-board meter). For 2021 PHEVs, fuel consumption was 240% above the test value according to the European Environment Agency (EEA), and for 2022 PHEVs, 306%.
• Fraunhofer ISI and the ICCT found real-world consumption about 3 times above the test value for private cars and about 5 times for company cars.
• In June 2026, the ICCT published a study of about 8 million new cars from 2021 to 2023, measured with the on-board meter. The real-world CO2 emissions of 2023 PHEVs are on average about 400% above the official value, so about five times as much. For combustion-engine cars, non-plug-in hybrids included, that gap is about 20%.
• The European Commission adjusted the calculation rule in 2025 and plans a further adjustment in 2027. According to the ICCT, the estimate is still wrong after 2025. A first draft report in the European Parliament wants to freeze further corrections.

With this real-world consumption, a PHEV comes to 163 grams per km. That is more than twice as much as an electric car and about 30% less than a petrol car. The ICCT uses the gap for 2021 PHEVs (+240%). The June 2026 study finds a larger gap for 2023. The IEA assumes 60% electric driving worldwide. A PHEV then emits about 40% more than an electric car. The result depends strongly on the share of electric driving, and so on how you use the car.

In 2020, Fraunhofer ISI calculated for German private drivers that a PHEV with about 90 km of electric range, charged daily, can drive fully electric on about 85% of driving days. That is about 70% of the kilometres.

From power plant to wheel: hydrogen and e-fuels

Every conversion costs energy. The think tank Agora Energiewende mapped the chain, based on a 2017 study by acatech. Grid losses count, because the chain starts at the power station:

Electrolysis uses electricity to split water into hydrogen and oxygen. A fuel cell turns hydrogen back into electricity. The percentages depend on assumptions, such as 70% efficiency for electrolysis. In the chain from Agora and acatech, a hydrogen car running on green hydrogen takes almost three times as much electricity as an electric car (69 against 26 per cent), and a car running on e-fuel about five times as much (69 against 13 per cent); the ICCT (2025) arrives at about 3 and 6 times.

The chart follows the electricity step by step to the wheel. Look at which steps cost the hydrogen and e-fuel routes the most.

From 100 kWh of electricity to the wheels

How much of 100 kWh of renewable electricity actually reaches the wheel, by type of car.

• Battery car
• Hydrogen car (green hydrogen)
• E-fuel car (combustion engine)
• Lost

For the same distance, a hydrogen car needs about 2.7 times as much electricity as a battery car, and an e-fuel car about 5.5 times as much.

Totals differ between studies because each study picks different step efficiencies and different boundaries.

Other studies: how much electricity reaches the wheel. Each study measures in its own way.

Study	Measure	Battery	Hydrogen	E-fuel
Volkswagen (2019) company	% of the electricity at the wheel	70 to 80%	25 to 35%	10 to 20%
Transport & Environment, chart (2017) advocacy group not checked against the source	% of the electricity at the wheel	73%	22%	13%
Transport & Environment (2020) advocacy group	% of the electricity at the wheel	77%	33%	16 to 20%
Ulf Bossel (2006) academic	kWh at the wheel per 100 kWh	69 kWh	19 to 23 kWh	-
Sterner, FENES, via DW (2017) press not checked against the source	range, battery car = 100	100	35	15 to 20
ICCT (2025) advocacy group	electricity use, battery car = 1	1×	3×	6×

Sources

• acatech, Leopoldina, Akademienunion, Coupling the different energy sectors: Options for the next phase of the energy transition (opens in a new tab) (2018 ) academic
• Agora Verkehrswende, Agora Energiewende, Individual and overall efficiencies for cars with different drive technologies, starting from renewable electricity (opens in a new tab) (2020 ) advocacy group
• Volkswagen AG, Hydrogen or battery? A clear case, until further notice (archived copy) (opens in a new tab) (2019 ) company
• Cerulogy (Malins) for Transport & Environment, What role is there for electrofuel technologies in European transport low carbon future? (opens in a new tab) (2017 ) advocacy group
• InsideEVs (Mark Kane), Efficiency compared: battery-electric 73%, hydrogen 22%, ICE 13% (opens in a new tab) (2017 ) press
• Transport & Environment, Electrofuels? Yes, we can ... if we are efficient (opens in a new tab) (2020 ) advocacy group
• Ulf Bossel, Does a Hydrogen Economy Make Sense? Proceedings of the IEEE 94(10) (opens in a new tab) (2006 ) academic
• Deutsche Welle, infographics crediting Sterner (FENES, OTH Regensburg) (2017 ) press not checked against the source
• ICCT, Life-cycle greenhouse gas emissions from passenger cars in the European Union: a 2025 update (opens in a new tab) (2025 ) advocacy group

Our own diagram, based on data from acatech and Agora (the three chains) and the studies in the table.

Show the figures as a table

Steps in each chain: efficiency and what is left of 100 kWh

Chain	Step	Efficiency	Left over (kWh)
Battery car	Power grid	95%	95
Battery car	Charging and discharging the battery	90%	85.5
Battery car	Electric motor	85%	72.7
Battery car	Mechanics (drivetrain)	95%	69
Hydrogen car (green hydrogen)	Power grid	95%	95
Hydrogen car (green hydrogen)	Electrolysis (making hydrogen)	70%	66.5
Hydrogen car (green hydrogen)	Compressing and moving hydrogen	80%	53.2
Hydrogen car (green hydrogen)	Fuel cell	60%	31.9
Hydrogen car (green hydrogen)	Electric motor	85%	27.1
Hydrogen car (green hydrogen)	Mechanics (drivetrain)	95%	25.8
E-fuel car (combustion engine)	Power grid	95%	95
E-fuel car (combustion engine)	Electrolysis (making hydrogen)	70%	66.5
E-fuel car (combustion engine)	Turning electricity into liquid fuel	70%	46.6
E-fuel car (combustion engine)	Long-distance transport	95%	44.2
E-fuel car (combustion engine)	Combustion engine	30%	13.3
E-fuel car (combustion engine)	Mechanics (drivetrain)	95%	12.6

Hydrogen is not always green:

• Grey: from natural gas, with CO2 emissions. In 2023, this was about 90% of all hydrogen worldwide (ICCT, after the European Hydrogen Observatory).
• Blue: from natural gas, with CO2 capture.
• Green: from water, with renewable electricity. Hydrogen from electrolysis, using any electricity, was less than 0.4% of world production in 2023.

The ICCT assumes 13.6 kg CO2e per kilo of grey hydrogen and 1.2 kg per kilo of green. In practice, a hydrogen car uses about 1.0 kg per 100 km.

On 30 October 2026, the H2Benelux dashboard showed eight public hydrogen stations in Belgium: two in Antwerp and one each in Erpe-Mere, Haasrode, Halle, Herve, Ollignies and Zaventem. No station is in Limburg. The trade press reported a price of 9.99 euros per kilo at DATS 24 in September and December 2025. In the Netherlands, a kilo cost 15 to 25 euros (RVO, the Netherlands Enterprise Agency, second quarter 2025). No more recent or official price could be found. These are indications, not fixed tariffs.

Worked example (illustrative)
At 1.0 kg per 100 km and 10 to 25 euros per kilo, 100 km of hydrogen costs 10 to 25 euros. An electric car uses 20.2 kWh per 100 km. At an assumed 0.30 euro per kWh at home, that is 20.2 × 0.30 = about 6 euros.

And the air we breathe?

An electric car has no exhaust. According to IRCELINE, the Belgian air quality institute, this removes emissions of NOx, carbon monoxide and volatile organic compounds. Under the Euro 6 standard, diesel cars may emit more NOx than petrol cars: 80 against 60 mg per km. For the mass of fine particles from the exhaust, the standard is the same: 5 mg per km. These are standards, not measurements.

Fine particles also come from tyres, brakes and the road surface, for every car. Exhaust standards have become stricter. As a result, wear is now the main source of fine particles for cars with a combustion engine. Electric cars are on average heavier. IRCELINE therefore expects more wear. Regenerative braking (the motor brakes and charges the battery) spares the brake discs. IRCELINE writes that the two effects roughly cancel out. The total of fine particles is then of the same order of magnitude as for a combustion-engine car. The net effect depends on the type of car, the tyres and your driving style.

For more on mining and raw materials, read the article "Mining and oil: what electric and fossil-fuel cars take from the ground".

What now?

1. Calculate with real-world consumption, not the brochure. Add 20% to the test consumption of a petrol or diesel car and 25% to that of an electric car, as the ICCT does.
2. Consider a plug-in hybrid only if you can charge every day, at home or at work. For how to charge at home, read the article "Electric driving and home charging".
3. Compare models with the Ecoscore of the Vlaamse overheid (Flemish government, ecoscore.be). It combines greenhouse gases, air pollution and noise.
4. Apply the worked example above to your own electricity. The cleaner your electricity, the sooner the electric car makes up its extra emissions.

Sources and disclaimer

This text is based on:

• ICCT, Life-cycle greenhouse gas emissions from passenger cars in the European Union (July 2025).
• ICCT, press release on plug-in hybrids (2 June 2026)
• IEA, Global EV Outlook 2026: Do electric cars reduce emissions overall?
• Fraunhofer ISI and ICCT, Real-world usage of plug-in hybrid electric vehicles (2020).
• Fraunhofer ISI and ICCT, Real-world usage of plug-in hybrid vehicles in Europe: a 2022 update (fact sheet, June 2022)
• Elia Open Data: CO2 intensity of Belgian electricity
• Car-Pass, Annual Report 2025. Yearly kilometres of Belgian cars.
• ICCT, From laboratory to road: a 2018 update of official and real-world fuel consumption and CO2 values for passenger cars in Europe. Gap between test and practice.
• ICCT, European Vehicle Market Statistics 2019/20. Official emissions of new cars by fuel.
• Carbon Brief, Factcheck: 21 misleading myths about electric vehicles (24 October 2023). Press, not a primary source. Its chart on old cars applies to the UK.
• IRCELINE: does an electric car still cause air pollution?
• Agora Energiewende: efficiency of drive routes (after acatech, 2017). Think tank, older study.
• H2Benelux: hydrogen stations
• Mobility Energy: hydrogen prices (September 2025) and December 2025. Trade press, not a primary source.
• DieselNet: EU emission standards for passenger cars. Technical reference that summarises the European regulations, not a primary source.

This is general information, not personal advice. Figures and rules may change. As of 30 October 2026.