How we test cars

Our lab and road tests allow us to thoroughly assess each car and give our verdict on all the things that really matter to car owners.

From overall handling to how accurate manufacturer fuel efficiency or electric car range claims really are, we put every car through its paces.

Plus, our annual reliability survey enables us to incorporate feedback from tens of thousands of car owners. This means we can tell you what a car's like to live with and reveal the most (and least) reliable cars.

Whatever your needs, you can trust our unique independent testing to tell you everything you need to know when choosing your next car.

Find out which cars excelled in our tests in our round-up of the best cars for 2025.

Which? car testing

Along with detailed reliability information, each car review includes ratings on a number of lab and road tests, allowing us to assess how it performs against its peers.

Click the links below to find out more about each set of tests.

• Reliability
• Safety
• Efficiency
• Emissions
• Comfort and practicality
• Ease of driving
• Performance

Best electric cars for 2025: discover the EVs that aced our tough lab tests

Reliability

Every year, tens of thousands of people tell us about the cars they own in the Which? car survey.

Through this survey we learn about a car’s common faults and overall owner satisfaction. Based on the results, we award reliability scores covering three age groups (0-4, 5-9 and 10-15 years). 

The information you give us also helps us track any issues, such as potential problems with specific models.

Model reliability

In our most recent Which? Car Survey, we had enough responses to generate reliability scores for 367 different car models, with results for 523 models across the three age groups we survey. Any car that we know to be unreliable (with a reliability rating of two stars or less) can't be a Best Buy.

Brand reliability

We assess the overall reliability of car brands through our survey and also publish data on how reliable each brand’s cars are at a fuel-type level. We had enough responses in our latest survey to give reliability scores for all of the major brands currently active in the UK. 

Safety

At Which? we take safety seriously. That’s why we factor a car’s Euro NCAP score (when it has been tested by the crash safety organisation) into our scores for each car we test. We also conduct our own hazard-avoidance and stability tests to identify cars that may leave you in a dangerous situation for longer in the event of an emergency.

Euro NCAP

We assess how well adult and child occupants, and pedestrians, are protected in a crash, by looking at what active safety systems a car has. This includes lane-keep assist (which should keep you in the intended lane), blind-spot assist (which lets you know when someone is in your blind spot) and autonomous emergency braking (AEB). We also look at passive safety systems, such as how many airbags there are.

All cars tested by Euro NCAP are given an overall star rating out of five. If a car that has been tested by Euro NCAP gets two stars or less, we make it a Don't Buy car, as we believe modern cars should earn at least three stars or more to be considered safe. Only cars that score four stars or more are eligible to be a Best Buy.

Since 1997, Euro NCAP has used crash-test dummies to record how well a car's occupants are protected in a crash, should the worst happen. But in more recent years, as passenger safety systems have become more advanced, the test also looks at the effectiveness of active safety systems designed to prevent a collision from happening in the first place.

Active safety systems monitor the environment around the car and are designed to take preventive action to avoid an impact, or at least mitigate the worst effects.

Some cars have safety kit that you can purchase as an optional add-on or with higher trim versions of the car. In these cases, Euro NCAP awards the vehicle two separate star ratings – one with the optional safety kit, and one without.

When two star ratings are available for the car, we always use the lower safety rating from the version of the car without the additional safety equipment. This is because we believe that safety shouldn't be an optional extra.

Which? safety assessments

We factor in the safety equipment that’s factory-fitted or optional, along with how well the car protects its occupants and pedestrians in the event of an accident and how easy it is to fit a variety of child car seats.

We also asses cars based on other safety features. This includes the 'pinch' test, which measures the force it takes to stop a car window from closing.

This safety measure should prevent damage to objects (such as fingers), but our testing reveals some cars are better than others.

Efficiency

Many cars struggle to meet their official fuel economy (mpg) figures when they’re faced with our tougher - but more realistic - tests, which include a motorway element to truly assess fuel consumption at high speeds.

We also measure emissions in the same stringent tests and have found a large number of cars that, although they pass official tests, emit big quantities of toxic gases such as NOx and CO. 

Petrol, diesel and mild-hybrid efficiency

Our fuel economy test is carried out on a rolling road in a lab. This simulates wind resistance and enables the test to be repeated in the same way in the exact same environment, as we use the same lab for each car test.

We use the Worldwide-harmonised Light vehicles Test Cycle (WLTC) cycle and run both a ‘cold cycle’ (where the engine is cold) and a ‘hot cycle’ (where the engine is already warm).

We also have a motorway driving cycle, where we accelerate up to – and sustain – motorway speeds. This cycle catches out a lot of cars that are efficient at low speeds but burn a lot of fuel on the motorway.

We also set the conditions of the test to be as realistic as possible:

• We test cars in the default setting they start up in, rather than switching to a more economical driving mode.
• Our additional test cycles are conducted with the air conditioning on, lights on dipped beam (or daytime running lights) and the radio switched on. Our trained drivers also stick to the test’s speed restrictions (with a tolerance of 2kph).
• We don’t tamper with the cars by removing extra weight or inflating the tyres any more than specified in the owner’s manual.

Full hybrid efficiency

When testing the efficiency of full hybrids (also known as self-charging hybrids), we conduct our test with a 60-70% state of charge displayed on the on-board computer.

Plug-in hybrid efficiency

For plug-in hybrids, we start our motorway and warm cycles with an empty battery. We then run our cold cycle immediately afterwards, with whatever battery charge the car has after completing the warm cycle. We also run a custom cycle with a fully charged battery, which we repeat until the engine turns on. After this cycle, we charge the car to work out how much electricity has been used. The ratio between pure electric and engine-only driving of each car accounts for how much we take each type of driving into account with our consumption figures.

Electric car efficiency

We test electric car efficiency slightly differently to other fuel types, using a test programme that combines the WLTC cycle and our own motorway cycle.

Before the test, the car is charged to full capacity and, just like other vehicles, the air conditioning is set to 20°C and daytime running lights (or low beams) are on throughout the cycle.

Through repeat tests, we're able to work out a realistic range that an electric car can travel.

Emissions

While testing a car's efficiency, we also collect air-polluting emissions data. This is the quantity of exhaust emissions – such as oxides of nitrogen (NOx), carbon monoxide (CO), hydrocarbons (HC) and particulate matter (PM).

Any vehicle that produces a very large amount of air pollution in our tests can't be a Best Buy. 

Volkswagen was caught out in September 2015 by employing a ‘defeat device’ in its diesel cars to dupe US tests over NOx emissions. This became known as the VW emissions scandal, or ‘dieselgate’. In Europe, 8.5m cars had the same device.

Whenever we're suspicious of the NOx emissions we record in the lab, we now run additional test cycles using a Portable Emissions Measuring System (PEMS). This will expose any car that switches to a more economical running mode if it detects it's in a lab.

The PEMS device is placed in the car, and the emissions are recorded while driving on real roads. We then compare this NOx emissions data with what we recorded in the lab and penalise any cars that have an incredibly high real-life reading.

While this is a realistic way of testing cars, we couldn’t replace our lab testing with PEMS testing alone. Being outside the lab, we aren't able to control factors such as temperature or road conditions that can affect the result.

To find out which cars are the most – and least – polluting, use our low-emission car checker.

How we rate car emissions

Comfort and practicality

Comfort and practicality includes how easy it is - or not - to get into and out of the car, and the maximum height of occupants. We also measure road, wind and engine noise and assess how well the car deals with bumps from large potholes to jittery surfaces like cobblestones.

Entry and exit

We assess every car on how effortless it is to enter and exit. This is particularly important if you or any of your regular passengers have mobility issues or are particularly tall.

This means you'll know if it’s a case of just sliding in easily, or whether you’ll end up doing an awkward side shimmy or bending over double just to get yourself into the vehicle.

Seat comfort

We also look at how well padded and supportive the seats are, making expert assessments of lumbar, thigh and head support. Our tests reveal which cars will keep you comfortable on long road trips and which won’t.

Ride quality

How comfortable you’ll be depends on how well the car handles the lumps, bumps and gaping potholes that are common on UK roads. Our test track has all those features and more. We drive over them repeatedly to assess a car’s suspension.

Quietness

We both measure and subjectively assess the trio of road, wind and engine noise in each of our tested models. If a car is noisy the entire time you’re driving, you won’t want to be inside it for long.

Seat space

We don’t just get inside the car and comment on the amount of room inside. We also use a dummy to work out precisely how much head, leg and knee space there is for the driver and all passengers.

Boot space

Car manufacturers vary in the way they measure their claimed boot space, and can include removing carpets or counting spaces that would normally be filled with a spare tyre. 

We load the boot with foam blocks to figure out the usable amount of space, ignoring useless nooks and crannies.

As well as providing you with the exact measurement of each boot, we give each car a boot space rating out of five, so you can easily compare cars at a glance.

Ease of driving

To find out how easy a car is to drive, we assess a number of areas from visibility and handling to the layout of the cabin and how smoothly the engine delivers its power.

Visibility

Being able to see out of your car and judge its dimensions easily is critical to safety and ease of use. Regardless of how many parking aids you've got fitted, a car with large blind spots will feel cumbersome to drive. Not only will it be more hassle to park and manoeuvre on tight city streets, you're more likely to miss objects or people close to the car while you're driving.

Below is the rear section of a 360-degree interior image of the Mercedes-Benz GLC Coupé (2016-2022). Obstructions to the driver’s all-round field of vision are colour coded depending on what the obstruction is and how severely it interrupts the driver’s vision. Rearward viability is particularly important, so the rear windscreen is split into two.

Dark blue: Anything that disturbs the view directly below the mid-point line is highlighted in dark blue.

Pink: Anything above the middle line directly to the rear is coloured in pink and is more detrimental to the visibility rating.

Light blue: This measures narrow windows. They are positive rather than negative, and improve the ratings for the adjacent pillars (coloured in yellow and pink).

Using this method we can get consistent and comparable data on all-round visibility for every car we test.

Smoothness of power delivery

If a car delivers its power in a smooth, refined and even manner, we’ll reward it for that. But if the power is delivered unevenly, the car’s engine vibrates a lot and the engine is very noisy when accelerating, then the car’s power delivery score will be poor.

 Driving stability

We put all cars through the Which? hazard-avoidance test. If you find yourself in a situation where you have to swerve around something, or someone, at the last moment, you need to be sure your car will remain controllable.

Travelling at speeds of up to 56mph, we conduct a test in which we swerve around an object and rejoin the original lane. This determines how stable and controllable a car remains, and whether it's likely to understeer (lose grip from the front and continue in a straight line) or oversteer (where the rear axle loses traction, potentially resulting in a spin).

The assessment takes into account the behaviour of a car's ESC (electronic stability control) system, and how likely it is to keep the car on its intended line, should you need to make a sudden manoeuvre.

Cabin layout

Our lab tests comprehensively assess how intuitive the car’s cabin controls are. We reward cars that have logically designed cabins, and we assess how easy it would be for someone to get used to the car’s interior controls. We also check if a car has poor interior ergonomics and reduce its test score in this category accordingly.

Performance

In-gear acceleration

We look at in-gear acceleration and see how well a car accelerates from 37mph to 62mph. This simulates moving to a faster lane on a dual carriageway or motorway, or overtaking a slow-moving vehicle on a country road. 

Brakes

In an emergency, you need to know if you’ll be able to brake quickly and without the car becoming unstable. To assess this, we conduct repeated emergency stop tests from 62mph (100kmph) and record the average distance covered over 10 attempts. Not only does this allow us to assess a car’s overall braking performance, but whether that performance is consistent and how likely it is that the brakes will overheat or fade. We score braking distances on a sliding scale, with 30 metres representing a perfect score, and 43 metres the very worst.

Drive smarter and cut costs using our expert advice. Get our Cars newsletter – it's free monthly

All images:  Which?/Abgedreht