What the MOT actually checks

What is and isn’t tested

The single most useful thing to understand about the MOT is how narrow it is. It answers one question — is this car roadworthy today — and it answers it thoroughly. Everything outside that question is invisible to it.

A tester works through a fixed list. Brakes, steering, suspension, tyres and wheels, lights and electrical equipment, visibility, body and structure, exhaust and emissions, seat belts, registration plates and a few items of other equipment. Each is inspected against published criteria and recorded as a pass, an advisory or a defect.

What the tester does not do is take anything apart. No panels come off, no fluids are sampled, nothing is dismantled. The inspection is visual and functional, carried out with the car on a ramp, in roughly forty-five minutes. That constraint is what defines the boundary between what the MOT sees and what it cannot.

Where the test actually finds things

Not every section produces the same amount of work. Across the national record, the sections dealing with what holds the car up and what slows it down account for the overwhelming majority of everything ever written on a certificate.

Counts are advisories and defects recorded in each section across the DVSA MOT testing dataset. They reflect how much a section generates rather than how serious it is — tyres are recorded constantly because every car has four of them and they wear every mile. Contains public sector information licensed under the Open Government Licence v3.0.

What the MOT never looks at

This is the list that costs people money, because a clean certificate says nothing about any of it.

When your car needs testing

A car faces its first MOT three years after it was first registered, and then every twelve months for the rest of its life. There is no upper age at which testing stops — a forty-year-old car in daily use is tested annually like any other, although vehicles built more than forty years ago and not substantially changed can be exempt.

Some vehicles are tested from one year old instead. Taxis, private hire vehicles, ambulances and anything built to carry more than eight passengers face their first test far earlier, on the reasonable basis that they cover far more miles carrying other people. If your car is used privately, the three-year rule is the one that applies.

You can test up to a month minus a day before expiry and keep your renewal date. That window is worth using: it gives you weeks rather than days to deal with anything that fails, and it costs nothing. The full rules on dates, retests and the fee are in section 15.

Common questions

Advisory, minor, major, dangerous

Five things can happen to any item on the test, and only two of them fail the car. The difference between the other three is where most of the confusion lives — and it is worth ten minutes, because two of them look identical on a certificate and mean quite different things.

A car passes with any number of advisories and minor defects. It fails on a single major or dangerous defect. The certificate lists each item under its own heading, and the wording differs: advisories say monitor and repair if necessary, minors say repair as soon as possible.

The one people get wrong: advisory is not minor

An advisory is not a defect at all. It is the tester recording something that is currently acceptable but heading in the wrong direction — a pad with a few millimetres left, a spring with surface corrosion, a tyre approaching the limit. Nothing about it is a requirement, and nothing happens if you ignore it.

A minor defect is a defect. It genuinely fails the standard for that item, but the failure is not judged to affect safety or the environment enough to fail the car. A number plate lamp out on a plate with two of them is the classic example: the standard is not met, the car still passes, and the certificate says repair as soon as possible.

Both let you drive away, which is why they get conflated. The practical difference is what they tell you. An advisory is a forecast — this will need doing at some point. A minor is a statement of fact — this is already wrong, and it was only the severity threshold that saved you.

What actually happens to an advisory

This is where our own analysis comes in, and it is the part nobody else has published. We took the national testing record and followed individual advisories forward to the same vehicle’s later tests, to see which ones came back as failures and which quietly went away.

From our analysis of 145,103,213 advisories tracked across 216,057,619 pairs of consecutive tests on the same vehicle. The three outcome columns account for every tracked advisory, so each row sums to 100%. Figures are by DVSA defect category rather than by individual part — code 5.3 covers the whole suspension, so a category figure is an upper bound for any single component within it. Contains public sector information licensed under the Open Government Licence v3.0.

Most advisories never fail anything. Even suspension, the worst of them, comes back as a failure fewer than one time in five. An oil leak comes back once in seventy-seven. So the honest answer to “do I have to fix this” is usually no — and the useful answer is that it depends enormously which advisory you were given.

And the gap is always a year

For the advisories that do become failures, the timing is remarkable. The median gap is 376 days for suspension, 378 for bearings, 379 for brakes, tyres, lamps, exhausts and oil leaks, and 384 for structural corrosion. Eight categories, eight days of spread.

Those components have nothing in common. A spring corrodes over winters; a bulb fails in an instant; brake fluid degrades on a calendar whether the car moves or not. If the figure were measuring how parts deteriorate, the medians would scatter across months. They do not scatter at all — which means it is measuring when owners act, and that is the next test. Section 12 takes that apart properly.

Common questions

Brakes

Brakes are inspected twice: once by eye and hand for condition, and once on rollers for performance. The performance half is where the test has hard numbers, and they are the most useful figures on this whole page — because unlike almost everything else, they are pass or fail arithmetic rather than judgement.

The condition check

The tester looks at everything visible without dismantling anything. Pads and linings where they can be seen, discs and drums from the outside, flexible hoses, rigid pipes, the master cylinder and servo, cables and linkages, and the fluid level through a transparent reservoir. Reservoir caps are not removed.

Categories from the DVSA MOT inspection manual, section 1. A vehicle passes with minor defects and fails on any major or dangerous one. Contains public sector information licensed under the Open Government Licence v3.0.

The performance test, and the numbers behind it

The car is driven onto rollers and each wheel is braked in turn. The tester records the force at every wheel, and the system works out two things: how much of the car’s weight the brakes can hold back, and whether the two sides of each axle match.

Efficiency is the headline figure. A car first used from September 2010 must reach 58% on the service brake. Anything older needs 50%, as do vans up to three and a half tonnes. The parking brake has its own, much lower requirement of 16% — it only has to hold the car, not stop it.

Falling short is a fail; falling far short is worse. Below the requirement is a major defect. Below half the required value — under 29% on a modern car — is recorded as dangerous, and the car does not leave under its own power.

Where more than half the wheels on a system lock on the rollers, the efficiency requirement is treated as met. Some four-wheel-drive cars cannot be roller-tested and are measured with a decelerometer on the road, or held on a gradient of up to 16% for the parking brake instead.

Imbalance: the fault you feel before you see

Both sides of an axle have to brake evenly. If one wheel produces less than 70% of the force of the other on the same axle, that is a major defect — a difference of more than 30%. On a steered axle, if one side drops below half the other, it is dangerous.

This is what a car pulling under braking actually is. The usual causes are a seized caliper slide, a partially collapsed hose acting as a one-way valve, or contaminated friction material on one side. All three are repairable without replacing everything, which is worth knowing before agreeing to a full set.

A binding brake shows up here too. If the rollers record significant effort with no brake applied, that is a major defect — and on the road it means a wheel that runs hot, kills fuel economy and destroys the pads on that corner.

What happens to brake advisories afterwards

Brakes generate an enormous amount of paperwork. Across the national record the section accounts for 158.7 million recorded items — second only to tyres and suspension — and it warns far more often than it fails, at roughly four advisories per defect.

That ignore rate is the strange one. Brakes are the system most drivers would name first if you asked what matters on a car, and the threshold is public and precise — 1.5mm, printed in the manual. Yet more than a third of brake advisories are still there when the car comes back.

The explanation is access, not indifference. Checking a tyre takes a coin and thirty seconds. Checking pad thickness properly means getting a wheel off, which most people cannot do on the street. The knowledge is public; the view is not. Tyres, which anyone can inspect while walking to the door, are ignored less often despite failing at a similar rate.

What it costs, and what to ask

Common questions

Tyres and wheels

Tyres are the most examined thing on any car. A single defect code — a tyre worn close to the legal limit — accounts for 161 million entries in the national record, more than any other item on any part of the test. They are also the one place where falling short is not merely a fail but a dangerous defect.

The tread rule, exactly as written

The requirement is more specific than most drivers realise, and in one respect more generous. The primary grooves must be at least 1.6mm deep across the central three-quarters of the breadth of the tread, and around the entire circumference. Both halves of that sentence matter.

Around the entire circumference is the half people forget. A tyre can measure comfortably above the limit at the point you happen to look and be illegal a few inches further round. Wear is rarely even, so checking one spot proves nothing — roll the car forward half a turn and look again.

The outer eighths being allowed to go bald surprises people, and it should not be read as permission. A tyre worn like that is reporting a geometry or suspension problem, and the replacement it needs is a symptom rather than the cure. It is also the pattern that reaches the legal zone next.

Everything else the tester looks for

Categories from the DVSA MOT inspection manual, section 5.2. Two things commonly assumed to be failures are not: an asymmetric tyre fitted the wrong way round is not a reason for rejection, only for advice; and a tyre pressure warning lamp lit because a pressure is low, rather than because the system has failed, should be recorded as advisory information. Contains public sector information licensed under the Open Government Licence v3.0.

The spare is not tested

Only the wheels and tyres fitted to the car at the time of inspection are examined. If a tester notices a defect on a spare they will mention it, but nothing about the spare affects the result. A car can pass with a spare that is a decade old, flat, or illegal.

Which matters more than it sounds, because the spare is what you fit at the roadside. Space-saver tyres have their own speed and distance restrictions printed on them, and a full-size spare that has sat under the boot floor for years may be well past the point anyone would fit it deliberately.

What the record shows about tyres

Risk rises with age and then falls back. Tyre advisories convert at 7.2% on a three-year-old car, 10.9% at ten years and 12.1% at fifteen — the peak. By twenty-six years the rate is down to 8.3%, which reflects which cars are still being presented rather than any improvement in tyres. Section 13 explains that pattern properly.

What to check yourself, and what it costs

Common questions

Suspension

Suspension holds two records in the national record, and they are the two you would least want. It is the advisory most likely to come back as a failure, at 18.7%, and the one least likely to be dealt with, at 44.2%. No other category on the certificate holds both extremes — and the reason has nothing to do with money.

How the tester finds it

This is the part of the test that needs the car off the ground. Wheels are jacked clear or placed on play detectors, and each one is levered, rocked and shaken while the tester watches the components. Where the front suspension has spring force acting on the lower arm, the car has to be jacked under that arm specifically, so the load comes off the ball joints and any wear actually shows.

Which is exactly why you cannot check this yourself. A worn joint under load looks and feels identical to a good one. The play only appears when the weight is off it — and that is the difference between suspension and every other item people are told to keep an eye on.

What fails, and what the phrase means

Categories from the DVSA MOT inspection manual, section 5.3. Two things commonly assumed to fail do not: a missing bump stop rubber is explicitly not a reason for rejection, and lateral play in a MacPherson strut top bush is normal on many cars when the suspension hangs free — it only counts if it is due to wear. Contains public sector information licensed under the Open Government Licence v3.0.

Misting is not leaking

A film of oil on a damper body is one of the most common things people are told needs replacing. The manual is unambiguous: slight seepage causing a film of fluid on a shock absorber is not a reason for rejection. Dampers are sealed against a moving shaft and a trace of oil past that seal is normal.

What does fail is severe leakage, or a damper that no longer damps. Those are different findings with a different look — running wet, a pool of oil, or a corner that keeps bouncing after a push. If the word on the certificate is misting or seepage rather than leaking, the car passed on that item — our guide to that phrase covers when it does become a fail.

What suspension advisories are actually about

Break the section into its parts and one component dominates everything else in the record.

The most common finding is also the cheapest fix. An anti-roll bar link is a short rod with a small ball joint at each end, and it is the usual cause of a knock over bumps. It is worth knowing that before a quote arrives, because nothing about the noise distinguishes it from an arm bush costing three times as much. Prices for the whole section are in our suspension repair cost guide.

The prescribed area, and why 30cm

Corrosion is judged differently depending on where it is. General body rust has to significantly reduce the structural rigidity of the car before it fails. But within 30 centimetres of any suspension, spring or sub-frame mounting, a much stricter standard applies — that zone is a prescribed area, and weakened structure there is a major defect on its own.

The logic is load. A mounting point carries everything the road does to the car, so metal near it is doing structural work that a wing or a door skin is not. It is also the reason a rust advisory near a mounting is a different kind of information from a rust advisory on a sill — and why nobody can price it until somebody has scraped it back.

The silence problem

Compare it with wheel bearings and the reason gives itself away. A bearing and a coil spring sit on the same corner of the car, cost roughly the same to replace, and are inspected in the same part of the test. Bearings get repaired 69.7% of the time. Springs and arms, 44.2%.

The difference is not money and not difficulty. A failing bearing hums, and the hum gets louder every week until somebody books it in. A corroded spring makes no sound at all, lights no lamp, and the car drives exactly as it did the day before it snaps. Owners act on warnings, and suspension is the one system that gives none.

Risk also peaks higher here than anywhere else. Suspension advisories convert at 7.3% on a three-year-old car and 22.3% at fifteen — the highest point of any curve in the record. Section 13 explains why the figure falls again after that, and why the fall says more about which cars are still being tested than about the cars themselves.

Common questions

Steering, axles and bearings

Steering is the section with the most precise published numbers in the entire test, and the least room for argument. It is also the section where warnings barely exist: it produces 7.4 million advisories against 6.2 million defects, a ratio of about 1.2 to 1, where suspension runs at four to one.

Free play: the clearest measurement on the test

The method is simple enough to picture. Wheels on the ground, pointing straight ahead, engine running if the car has power steering. The tester turns the wheel gently left and right as far as it goes without the road wheels moving, and measures how far a point on the rim travelled.

The distinction the manual insists on is worth carrying into any conversation with a garage. Play caused by wear or maladjustment fails. Play caused by the way the mechanism is built — a flexible coupling deflecting, a spring compressing in the power steering — does not. The same movement at the rim can be either, and telling them apart is the tester’s job rather than a measurement.

What else is inspected

Categories from the DVSA MOT inspection manual, section 2. Two findings commonly assumed to fail do not: a steering box that seeps oil is a minor defect and only fails when oil is actually dripping, and movement in a column with universal joints or a flexible top bearing is normal on many cars. Contains public sector information licensed under the Open Government Licence v3.0.

The gaiter, and why a minor here matters more than most

A split rack gaiter is a minor defect, so the car passes. That is the letter of it. What the grading does not convey is that the rubber concertina is the only thing keeping road grit and water off the rack’s internals, and a rack is one of the more expensive things under a car.

So this is the clearest case on the whole certificate for treating a minor as urgent. The gaiter itself is a modest job. Waiting until it becomes a major — missing, or no longer sealing — usually means the damage it was preventing has already started.

Wheel bearings: the one item judged by ear

Every wheel is raised, spun and listened to. The tester rocks it for play and spins it for roughness, and roughness is assessed by sound. It is the only part of the MOT where the verdict comes from hearing rather than seeing or measuring.

Which is exactly why bearings behave so differently in the data. Bearings and springs sit on the same corner of the car and cost roughly the same to put right, but 69.7% of bearing advisories are dealt with before the next test against 44.2% for suspension. A bearing announces itself — the hum gets louder every week — and owners act on noise.

That early peak is unusual and worth pausing on. Most categories climb to fifteen years or beyond. Steering tops out at ten, at 17.4%, and is down to 11.8% by twenty-one. Whatever wears out in a steering linkage does so on a shorter clock than springs, tyres or brakes — and the cars that survive past it appear to have had it dealt with.

What it costs

Common questions

Lights and electrical

Lighting is one of only two parts of the test where failures outnumber warnings — 35.7 million defects against 10.8 million advisories, a ratio of more than three to one. The reason is structural rather than harsh: a bulb works or it does not, and there is no gradual state for a tester to write down.

Headlamp aim: numbers, and no advisory band

Beam aim is one of the few things on the test with a published tolerance, and it is not a single figure. The car sits on a marked area, beam equipment is aligned to it, and the tester reads where the horizontal cut-off of the dipped beam falls against a set of lines. Where the acceptable band sits depends on how high the headlamp is mounted.

This is why an aim advisory is a warning rather than a grace period. Aim outside the band is a major defect and fails the car. If your certificate carries an advisory about beam aim, the beam was inside the band — close enough to the edge that the tester wrote it down anyway.

One detail worth knowing before you go in. The tester checks the beam without touching your dashboard levelling control. If it reads too low because the dial was left turned down, they re-check with it at its highest setting — but setting it back yourself removes the ambiguity, and costs nothing.

The modification that fails more cars than people expect

Everything else in the section

Categories from the DVSA MOT inspection manual, section 4. Two things commonly assumed to fail do not: sequential or dynamic indicators are explicitly not a reason for rejection, and on some cars with LED rear lamps a proportion of the stop lamp LEDs dim when the rear fog lamp is on, which is not a defect. Contains public sector information licensed under the Open Government Licence v3.0.

The number plate lamp, graded by how many you have

This one confuses people because the grading depends on the car rather than the fault. If your plate has two lamps and one has failed, it is a minor defect — the car passes, and the certificate says repair as soon as possible. If the plate has only one lamp and it is out, or if every lamp has gone, it is a major defect and the car fails.

It is also the item nobody can see from the driver’s seat. It sits behind you, it never dims as a warning, and it fails silently. A walk round the car with the lights on, once a month, removes it entirely — and the same walk covers every other bulb in this section. Our advisory decoder translates whatever wording ends up on your certificate.

Why lighting has almost no warning stage

A brake pad wears from eight millimetres to five to two, and somewhere in that range a tester can reasonably write that it is getting thin. A bulb has no equivalent. It illuminates or it does not, so there is nothing to warn about — which is why this section records three failures for every warning while suspension records one for every four.

The practical consequence is that the 376-day rule does not apply here. Everywhere else on the certificate, an advisory buys you roughly a year. In this section there is usually no advisory at all, so the only defence is looking before you book — the checklist in section 16 covers it in five minutes.

Common questions

Visibility

This is the most extreme section of the entire test. Across the national record it produces 11.7 million defects against 357,000 advisories — roughly thirty-three failures for every warning. Suspension runs at four warnings per failure. The difference between them is a factor of about 130, and it is not about severity.

Zone A, and where it actually sits

The tester sits in the driver’s seat and looks out. That is the whole method — field of vision through the windscreen, then through the front side windows, then the view of the mandatory external mirrors. What they are assessing is whether anything significantly affects the view of the road.

The detail people get wrong is what Zone A is centred on. It is a 290mm strip centred on the steering wheel, within the area the wipers sweep — not on the driver’s eyeline, and not adjusted for how tall you are. The manual says explicitly that the effect on tall or short drivers is not considered.

And the limits are not a switch. Failure for damage is only justified where it significantly affects the view of the road. If the only thing obscured is the sky or the bonnet, that is not a rejection. Ten and forty millimetres describe where the judgement usually lands, not a rule that fires on its own.

Wipers, washers and the bonnet catch

From the DVSA MOT inspection manual, section 3. The bonnet catch sits in this section for a reason that becomes obvious once stated: a bonnet that flies up while the car is moving is a visibility problem. Contains public sector information licensed under the Open Government Licence v3.0.

Why this section has almost no advisories

A brake pad goes from eight millimetres to five to two, and somewhere in that range a tester can reasonably write that it is wearing thin. A wiper blade has no equivalent. It clears the screen or it smears it. A washer bottle is full enough or it is not. A chip is inside the limit or over it.

Which means the warning stage barely exists here, and that changes what you can do about it. We hold no conversion figure for this section, because there are too few advisories to track — and that absence is the finding rather than a gap in the data.

So everything we have written about the 376-day rule is suspended in this section. Elsewhere on the certificate an advisory buys you roughly a year to plan. Here there is usually no advisory at all — which is why the five-minute check in section 16 matters more for visibility than for anything else on the car.

What to do before you book

Common questions

Body, structure and corrosion

Rust is the only thing on the whole test that behaves differently from everything else. Every wearing part climbs in risk to about fifteen years and then improves, because the neglected cars have left the road. Corrosion climbs, reaches a plateau, and never turns back down — and that single fact tells you more about buying an older car than any other figure we hold.

Two standards, and which one applies where

This is the distinction that decides most rust arguments, and the manual is precise about it. Section 6 covers the general structure and explicitly excludes prescribed areas — those are assessed under brakes, steering, suspension and seat belts instead. So the same patch of rust is judged by two different standards depending on what it is near.

A prescribed area is any load-bearing structure or supporting panel within 30cm of a mounting point for a tested item. The strictest standard applies there because that metal is carrying load. Elsewhere, the bar is much higher — the car’s overall rigidity has to be significantly reduced. Contains public sector information licensed under the Open Government Licence v3.0.

Why corrosion is the exception to everything else

We split every tracked advisory by the age of the car when it was written. Seven categories produce the same shape. Two do not.

The improvement in wear items is not the parts getting better. It is which cars are still being presented. A car that was being neglected at twelve is often gone by twenty — scrapped, exported or standing untested on a drive — so it stops contributing. What remains is a selected group of survivors.

Selection cannot rescue corrosion, and that is the whole point. A careful owner can postpone a clutch, nurse a bearing or replace a damper on their own schedule. Nobody can un-rust a car. So while the wear curves come back down, the rust curves climb to a level and stay there for every remaining year in the record.

And yet rust is the most repaired advisory of all

This looks like a contradiction and is not one. Structural rust converts to failure less often than brakes do, and gets fixed more often than anything else. What that combination measures is the weight of the consequence rather than the frequency of the fault.

A worn pad is a purchase. Structural rust is a decision about the car. When somebody hears that the metal near a mounting is going, they either commit to welding or they start planning the car’s exit — and both of those resolve the advisory. Very few people simply carry on, which is exactly what more than a third of owners do with brakes.

The rest of the section

Common questions

Emissions, exhaust and leaks

This is the section where the test has become dramatically stricter over time, and where the numbers are least known. The diesel smoke limit for a new car has fallen from 3.0 to 0.7 — a tightening of more than three quarters — and there is a precise, published figure for how big an oil puddle has to be before it fails.

Petrol: four readings, all of which must pass

The car is warmed up, the probe goes in the tailpipe, and the engine is held between 2,500 and 3,000rpm. Three readings are taken there, then a fourth once it settles back to idle. Every one has to be inside its limit.

Default limits for the basic emissions test. Older cars without a catalyst are tested to far looser figures — 3.5% CO and 1,200ppm HC — and some cars have manufacturer-specific limits that override the defaults. The manual also requires that a limit be exceeded continuously for at least five seconds before failing, because some engines give unstable readings. Contains public sector information licensed under the Open Government Licence v3.0.

Lambda is the one people have not heard of and the one worth understanding. It measures how close the air-fuel mixture is to ideal — exactly 1.00 is perfect, and the test allows three percent either way. A reading outside that window usually means a failing oxygen sensor, an air leak, or a fuelling fault, and it fails the car even if CO and HC are both fine.

Diesel: the limit that has fallen furthest

Diesels are not measured for gases at all. They are accelerated hard against a smoke meter, which records opacity — how much light the exhaust blocks, on a scale where zero is clear air.

The engine management light works both ways

The tester turns on the ignition and watches the malfunction indicator lamp. It should illuminate and then go out. If it stays on, the car’s own computer has found a fault and that is a major defect.

What catches people out is the other direction. A lamp that does not illuminate at all is also a major defect, because it suggests the bulb has been removed or the circuit tampered with. This applies to petrol cars first used from July 2003 and diesels from July 2008.

Fluid leaks: the 75-millimetre rule

There is a precise published figure for this and almost nobody knows it. With the engine idling, a leak fails if it creates a pool on the floor more than 75mm across within five minutes — or if several smaller leaks together drip at that rate.

Three fluids are explicitly excluded. Engine coolant, screen wash and the fluid used for selective catalytic reduction — AdBlue — are named in the manual as not being reasons for failure. So a coolant drip is a problem for your engine and not for your certificate.

That pair of figures is the most interesting on the whole page, because they belong together. Oil leaks are the most ignored advisory in the record and also the one that almost never returns as a failure. Those are not two findings — they are one.

Ignoring an oil leak advisory is, on this evidence, a reasonable decision. The wording testers use is usually “oil leak, but not excessive”, and the gap between not excessive and a 75mm puddle in five minutes is enormous. A weep from a cam cover can sit there for years without ever approaching the threshold — and the data says that is exactly what happens.

What it does not mean is that the leak is harmless to the car. Oil on a rubber bush degrades it, oil on a brake pad contaminates it, and a leak losing enough to matter will show on the dipstick. The MOT is telling you it is not an environmental or safety problem yet, which is a narrower statement than it looks. Our guide to that phrase covers where it typically comes from.

Exhaust noise, and what is not measured

There is no decibel limit in the MOT. The tester revs the engine to around 2,500rpm and judges whether the noise is unreasonably above what a similar car with a standard silencer in average condition would make. That is a judgement, deliberately, and it is why aftermarket exhausts sit in a grey area rather than a numbered one.

Emission control equipment is checked visually and only where visible. The manual limits this to components a tester can see and identify — catalytic converter, oxygen sensors, EGR valve, DPF, SCR. Missing, obviously modified or obviously defective is a major defect on cars first used from September 2002.

Common questions

What the MOT never checks

Everything above describes what a tester examines. This section is about the other list — and it is the more expensive one. The components most likely to cost you four figures are, almost without exception, the ones nothing in the test will ever mention.

The pattern behind the two lists

The test inspects what can be seen, measured or operated without taking anything apart. That is a sensible boundary for a forty-five minute roadworthiness check, and it produces a consistent side effect: tested items are mostly the cheap ones.

The cambelt, and the limit people miss

Nothing in the MOT looks at a cambelt, and it is the single most consequential omission. The belt drives the camshaft in time with the crankshaft; on many engines, if it breaks, the pistons meet the valves and the repair stops being a belt and becomes an engine.

Manufacturers publish two limits, and the second is the one that catches people. There is a mileage figure, which most owners know about, and an age figure, which most do not. Rubber degrades on a calendar whether the car moves or not — so a low-mileage car can be well past its belt without ever approaching the mileage number.

Wet belts add a further wrinkle. Some modern engines run the belt inside the engine, lubricated by oil. They degrade differently, and shed material that can block the oil pick-up — which turns a belt problem into an oil starvation problem. Our guides cover scheduled belts at £300 to £880 and wet belts at £450 to £1,000, with model-specific figures on the individual pages.

The four checks people assume are covered

And the rest of the untested list

National independent-garage ranges, parts and labour. These are estimates for planning rather than quotes — final prices come from garages, and vary considerably by model. Every one of these can be priced for your specific car through the reg check.

What this means when buying

A long clean MOT history is genuinely valuable and routinely over-read. It proves the car has been kept roadworthy and shows what a professional noticed each year, which is more than any seller’s description gives you. Our used car check reads that record and prices whatever it contains.

What it cannot do is tell you about the top of that chart. Ask when the cambelt was last changed and for the invoice. Drive the car properly and feel the clutch bite point. Look at the oil on the dipstick and the coolant in the expansion tank. None of that appears in any record, and all of it costs more than everything the MOT does cover.

Common questions

What happens after the test

Everything above this point describes the day of the test. This section is about the year that follows — and it contains the one finding from our analysis that surprised us enough to check the code twice.

How we measured it

An advisory is a prediction, so it can be scored. The tester is saying that something is heading in the wrong direction. Whether it actually did is a question the national record can answer, because every test is linked to the vehicle that took it.

We paired 216 million consecutive tests on the same vehicles and followed 145 million advisories forward to see what happened to that specific item. Three outcomes are possible: it came back as a failure, it was still sitting there unresolved, or it had been dealt with. For the ones that failed, we recorded how many days passed.

The result

Why this should not happen

These components have nothing in common. A coil spring corrodes across winters of salt and water. A bulb filament fails in an instant, from vibration and hours of use. Brake fluid absorbs moisture from the air on a calendar, whether the car moves or not. Tread wears by the mile. Rust does not care how far you drive.

If the figure were measuring deterioration, the medians would be scattered across months. Materials with completely different failure mechanisms, operating on completely different timescales, have no reason to reach a threshold at the same moment. Yet eight categories land within eight days of each other.

Two things the figure does not say

It does not mean most advisories fail. Conversion rates run from 18.7% down to 1.3%, so the overwhelming majority never become anything. The median describes only the ones that did convert — it is about when, not whether.

And it is a median, not a distribution. We know the midpoint; we do not know how many converted after thirty days or after seven hundred. Anyone quoting this as “advisories fail after a year” has read one number as though it were a rule.

Two sections are also outside it entirely. Lighting and visibility barely produce advisories at all — a bulb works or it does not — so there is no meaningful warning period to measure. Section 07 and section 08 cover why.

What to do with it

If the number is really about timing rather than parts, then it is a planning tool rather than a warning.

On an older car there is a fourth question, and it is the one that decides the rest. When the advisory list starts approaching a meaningful share of what the car is worth, the decision stops being about which repair and becomes one about the car. Our worth-fixing calculator does that arithmetic honestly, and our quote checker tells you whether a price you have been given is normal.

Common questions

How age changes the odds

Split every tracked advisory by how old the car was when it was written, and a pattern appears that nobody would predict from mechanics alone. Most categories climb to a peak somewhere between ten and eighteen years — and then get better. Understanding why is the difference between reading this data and being fooled by it.

Where each category peaks

The numbers behind it

Steering and lamps start later in this table because the youngest age bands did not clear our minimum of 30,000 advisories, not because those cars had none. National figures from our analysis of the DVSA MOT testing dataset. Contains public sector information licensed under the Open Government Licence v3.0.

Why the improvement is not an improvement

The obvious reading of a falling curve is that very old cars are somehow better. They are not. What changed is which cars are being measured. A car that was being neglected at twelve years old is often gone by twenty — scrapped, exported, or standing on a drive untested — and once it stops being presented, it stops contributing to the data entirely.

So the twenty-five-year-old cars in this record are a selected group. Keeping a car that long is a decision, and it tends to come with either real maintenance or very few miles. Read that way, the late figures are as much a portrait of the owners as of the vehicles.

A second caveat, and it is the one people miss

“Fifteen years old” pools different calendar years. The record spans two decades of testing, so a fifteen-year-old car measured early in the dataset was tested under a different standard, against a different fleet, from a fifteen-year-old car measured near the end of it.

What we have is a like-for-like comparison of vehicle ages, not of calendar years. That is still far better than comparing whole model histories, which is what most published failure rankings do — but it is not the same as a controlled experiment, and we would rather say so.

What to take from it

And on an older car there is a threshold worth naming. When a single repair approaches half of what the car is worth, the decision stops being about the fault and becomes one about the car. That is also the point at which cars start leaving this dataset — our worth-fixing calculator does that sum with real figures.

Common questions

Model, fuel and engine size

Age explains a great deal, but not everything. Hold age constant and three other variables still move the odds — which car it is, what it burns, and how big the engine is. The third of those produces the strangest result in the whole dataset.

Engine size: the U-shape nobody expects

The reasonable assumption is that bigger engines mean more to go wrong. Split failure rates by engine capacity and the opposite appears at both ends — the extremes do best, and the middle of the range does worst.

The most likely reading has nothing to do with engineering. A car with a 2.5-litre engine or larger was expensive when new and is usually still worth something, so the arithmetic of repair against value stays favourable for longer. Cars under a litre are mostly recent city cars — simple, lightly used, and not yet old enough to have accumulated much.

The middle of the range is where ordinary family cars live. High mileages, modest residual values and complicated enough to have plenty to go wrong. That combination, rather than the capacity itself, is the plausible explanation — and we would rather offer it as a reading than dress it up as a finding.

One detail supports that. The 2.0 to 2.4 band is the worst of all at three years old, at 18.8%, and then overtakes every other mid-range band by fifteen. That is the shape you would expect if the band contained a lot of diesels, which fail most often when young and then converge with petrol.

Fuel: the clearest ranking we have

Hybrids are lowest at every age we measured, and the gap does not close — a twelve-year-old hybrid fails less often than a six-year-old petrol. The electric figures rest on a much smaller and much earlier population, so treat them as an early signal rather than a settled result. The MOT does not test the battery, motor or power electronics on any electrified car. Contains public sector information licensed under the Open Government Licence v3.0.

LPG is the outlier at the wrong end, and it has not appeared anywhere else in our work. Converted cars fail more often than diesels at every age in the range — 18.8% at three years against 17.6% for diesel, and 31.3% at ten. The sample is small at 681,000 tests, but it is consistent across the whole span rather than noisy.

Which of the three matters most

The model is the biggest single lever, by some distance. Comparing 268 models at exactly ten years old, failure rates run from 14.6% to 39.2%. The same comparison at fifteen years, across 157 models, puts the same names at both ends — which is what makes it a signal rather than noise.

And what none of it tells you

These are populations, not predictions about a specific car. A well-kept example of a poorly-ranked model will outperform a neglected example of a well-ranked one every time, and the gap between two individual cars of the same model and age is far wider than the gap between models.

None of it measures reliability either. The MOT tests roadworthiness on one day and never inspects the clutch, gearbox, cambelt or engine internals. A model can pass the test consistently and still be expensive to own — those are different questions, and only one of them is in this data.

What is worth doing is reading the individual record. Our used car check reads a car’s own history and prices whatever it contains, and the worth-fixing calculator handles the question that eventually replaces all of these — whether this particular car is still worth the money.

Common questions

Dates, retests and the fee

The rules around the test are simpler than the test itself, and worth knowing because two of them save real money. The fee is capped by law, and a failure does not automatically mean paying twice.

The fee is a ceiling, not a price

£54.85 is the most a test centre may charge for a car, and most charge considerably less. It is a statutory maximum set by DVSA, unchanged for well over a decade, and it applies identically to a franchised dealer and a back-street garage. There is no VAT on it — a price quoted as the test fee “plus VAT” is wrong.

Cheap MOTs are usually a way of getting your car onto a ramp. A test advertised well below the cap is a loss leader that puts a technician in front of a vehicle with a fresh list of things it needs. That is not sinister, and it can be good value — but it is worth reading a failure sheet as a quotation rather than an instruction. Our MOT cost guide covers the full fee table by vehicle class.

What happens if it fails

Whether a retest costs anything depends almost entirely on where the car goes next, and on how quickly it comes back.

The next-working-day route only covers a specific list. DVSA names the items that qualify — lamps, wipers, tyres, seat belts, doors, mirrors, the fuel filler cap, registration plates and similar quick fixes. A failure on brakes or structure does not qualify, however fast you return.

Dates and the early window

The early window is the single most useful rule here. Using it costs nothing and buys up to four weeks to deal with a failure, which matters most when a part has to be ordered. Ignoring it means a failure discovered on the expiry date, with the car unable to be legally driven.

Driving without a valid certificate

There are exactly two exceptions and they are narrow. You may drive to a test you have already booked, and you may drive to a garage for repairs. Both require a direct route, and your insurance has to be valid — most policies are not valid without a current MOT, which is worth checking rather than assuming.

Everything else is an offence carrying a fine of up to £1,000. The certificate is not what makes a car legal to use; roadworthiness is, and the certificate is evidence of it on one particular day.

Common questions

Checklist before you book

Everything above explains what a tester does. This is the short version of what you can do first — and it is worth more than it sounds, because more than a quarter of every defect ever recorded sits in the two sections you can check on your own driveway in ten minutes.

Two things make that figure worth acting on rather than just noting. The items are cheap — bulbs, blades, washer fluid, a chip repair. And they are precisely the sections that give no advisory warning, as section 07 and section 08 set out. Everywhere else on the certificate you get roughly a year’s notice. Here you get none.

Ten minutes, no tools

Five more that take a minute each

Also worth doing before you set off: turn the headlight levelling dial back to its normal position if it has been nudged, and make sure the dashboard warning lamps illuminate and then go out when you turn the ignition on — a management light that never comes on fails just as one that stays on does.

What not to bother with

Washing the car makes no difference to the result. A tester may refuse to test a vehicle that is genuinely filthy underneath — because they cannot see what they are meant to inspect — but ordinary road dirt is not a factor, and a spotless car does not pass more easily.

Nor is it worth clearing the boot for the sake of it, beyond making sure the tester can reach anything they need to see. What does matter is being straightforward about anything you already know is wrong: hiding a fault wastes a test fee rather than avoiding one.

Timing the booking

Use the full early window. Testing up to a month minus a day before expiry keeps your renewal date and buys up to four weeks of slack. That is the difference between a failure being an inconvenience and being a car you cannot legally drive. The rules are in section 15.

If you are also having the car serviced, do it before the test rather than after. Trivial failures get caught first, and anything found underneath can be quoted while the car is still on the ramp. Our service cost guide covers what each level includes.

And read the certificate properly when you get it back. Failures are today’s problem; advisories are next year’s forecast, and they are the more useful half of the document. Our advisory decoder translates each phrase and prices it for your exact model.

Common questions

Methodology and citation

Every outcome figure on this page comes from one analysis, and it is worth being explicit about how it was done — partly so you can judge it, and partly because we would like people to use it.

What we did

The DVSA publishes the full MOT testing record under the Open Government Licence. It contains every test carried out, every item recorded, and the vehicle each belongs to. What it does not contain is any analysis of what happened next — and that is the gap this page fills.

We paired consecutive tests on the same vehicle and followed individual advisories forward. For each one we recorded which of three things happened by the following test: it became a failure, it was still there unresolved, or it had been dealt with. Where it became a failure, we recorded the number of days.

What it cannot tell you

These limits are real and we would rather state them than have them found.

Every question on this page

Sixty-four questions, answered in the section they belong to. Each one links to its answer.

Related guides