The claim is that cars built in the 1980s and 1990s ran to 300,000 or 400,000 kilometres without major work, and that modern ones do not. As a statement about the whole fleet, the registration record makes that hard to sustain. As a statement about the cars people actually remember, it is probably true - and that is the problem.
- Applies to
- Any used-car buying decision
- Best available evidence
- Same-age roadside and compulsory-inspection data
- US cars still registered at 20 years
- 9.16%, model years 1977-2003
- Australian by-age defect data
- No national series published
- Pre-2003 warranty data
- We could not find a public series
The cars you remember were the survivors
NHTSA's technical report on vehicle survivability (DOT HS 809 952, S. Lu, January 2006) tracked registration counts for model years 1977 to 2003. Of passenger cars, 94.13% were still registered at five years, 78.66% at ten, and 9.16% at twenty. Light trucks did better: 69.56% at ten years, 25.42% at twenty, 6.73% at thirty. Expected lifetime travel worked out at 152,137 miles for cars and 179,954 for light trucks - roughly 245,000 km for a car.
Read those numbers against the folk claim. The report puts the expected life of a car from exactly the era people describe as indestructible at about 245,000 km, and finds fewer than one in ten still registered at twenty years. The 400,000 km cars were real. They were a minority of their cohort, and the rest were deregistered and crushed before anyone could form a memory of them.
Two cautions on that data. Survival here means still registered, which is not the same as still roadworthy. And light trucks outlasting cars is its own trap: comparing a remembered 1990s ute against a modern hatchback compares two different survival curves, not two eras.
A comparison that holds the bias constant
Comparing a 1995 car to a 2025 car tells you little, because one has already passed thirty years of scrappage filtering and the other has passed none. The fix is to compare cars of the same age from different build eras: a ten-year-old car in 2015 against a ten-year-old car in 2025. Both have survived exactly ten years of the same filter, so what differs is the build era rather than the selection.
Germany's ADAC publishes that comparison from its roadside callout data. A ten-year-old car's annual breakdown probability fell from 6.5% in 2015 to 3.1% in 2025, and a five-year-old car's from 3.6% to 2.1%.
Keep the limits in view. This is the German fleet, and ADAC states its callouts cover roughly half of all breakdowns in Germany. Accidents, tyre damage and self-inflicted faults such as running out of fuel are excluded. Models need a registration threshold to be included at all, the current analysis covers vehicles aged two to ten, and ADAC itself notes that owner maintenance has more influence as vehicles age. It supports a statement about German cars in that age band. It does not support "cars worldwide are twice as reliable".
It is also worth knowing what actually stops them. In ADAC's 2024 data, across over 3.6 million callouts, the 12V starter battery accounted for 44.9% of all breakdowns - 44.6% for combustion cars and 50.5% for EVs, and the most common cause by a wide margin, well ahead of engine or transmission faults. A flat battery is not always simply a worn consumable: ADAC's 2025 breakdown report names model-specific component and charging-system defects, including the Hyundai IONIQ 5's charge-control unit. Australian figures may differ.
Age matters more than build era
Germany's Hauptuntersuchung is compulsory and near-universal, so its denominator is close to the entire registered fleet - which is why it is worth more than a voluntary survey. TÜV NORD's main inspections between June 2023 and June 2024 found 60.2% of vehicles completely defect-free, 14.2% with minor defects, 25.4% with significant defects requiring repair, and 0.06% unroadworthy and withdrawn immediately.
Broken out by age, the gradient is steep. Significant defects appeared on 6.2% of three-year-old vehicles and 9.1% of five-year-old ones. The TÜV Report 2026 records 28.1% of 12- to 13-year-old cars with major defects, and in the 10- to 11-year band puts Mercedes at 18.5%, Audi at 19.2% and Toyota at 22%.
Wear items - brakes, bushes, tyres, lighting, corrosion, exhaust - accumulate with time and distance regardless of how well a car was built. That gradient is why "my old car has been fine" and "old cars fail inspection more often" are both true at once. One is a survivor speaking. The other is a population being measured.
What the rules now require of an engine
Two regulatory facts sit awkwardly with the idea that manufacturers build engines to expire. US emissions certification requires a vehicle to still meet its standard at the end of a defined useful life, and that life has been ratcheted upward rather than relaxed: Tier 3 and LEV III set full useful life at 150,000 miles or 15 years for several vehicle classes, allow LDV and LDT1 to certify to 120,000 miles because the Clean Air Act precludes EPA from requiring longer, and eliminated the separate intermediate 50,000-mile standard. Separately, US federal law requires an 8 year / 80,000 mile warranty on specified major emission control components - the catalytic converter, the emissions control computer and the onboard diagnostic device - on top of a baseline 2 year / 24,000 mile emissions warranty.
Emissions compliance depends on fuelling accuracy, ignition, oil consumption, valve sealing, sensor function and catalyst function. An obligation to still comply at 150,000 miles therefore forces margin into the powertrain itself, not merely into the exhaust. What changed is where the margin lives: less of it is generous tolerance, and more of it is a specification the car has to be kept inside - an oil grade, a service interval, a duty cycle.
These are US figures. Australia's ADRs reference UN and European rules, so the mileages do not apply directly here. And we could not find any primary source supporting the popular claim that modern engines are deliberately engineered to fail once the warranty expires. The two obligations above cut against it.
Where this article is weakest
Something is genuinely getting more expensive. Ford accrued $6.71bn for warranty across 2025, up 7%, at an accrual rate between 3.9% and 4.5% of product revenue, while GM accrued $5.07bn, up 10%, with claims paid up 19%. Whatever is happening to mechanical durability, standing behind a modern vehicle is not getting cheaper. Read it carefully though: an accrual is a forward-looking estimate that absorbs expected recall campaigns and rising cost per repair, and Ford's claims actually paid fell 2% in the same year its accrual rose.
Owners are reporting more problems too. J.D. Power's 2025 U.S. Vehicle Dependability Study, which surveys original owners of 2022 model-year vehicles after three years, recorded an industry average of 202 problems per 100 vehicles, 12 PP100 worse than 2024 and the highest problem level since 2009. Software is a visible part of that: the study puts Android Auto and Apple CarPlay connectivity at 8.4 PP100, up from 6.3, and reports that half of the industry's top ten problems related to smartphone integration, usage or connectivity. But J.D. Power puts software defects at only 9% of all problems reported, across 184 problem areas in nine categories, and attributes the year's decline mainly to pandemic-era build conditions - models newly launched for 2022 averaged 241 PP100 against 196 for carryover models. It is US data, and measuring three-year-old cars against three-year-old cars says nothing about how a fifteen-year-old car compares.
German inspections have moved the same way: the overall defect rate has risen for four consecutive TÜV Reports to 21.5%, up 0.9 points against a low of 17.9% in 2022, which TÜV attributes substantially to record fleet age rather than build quality - but the raw direction is upward.
The commonly cited reply, that the average car on the road is older than it used to be, is weaker than it sounds. S&P Global Mobility put the US average light-vehicle age at 12.8 years in 2025, and its own analyst attributes the rise to elevated new and used prices and economic uncertainty rather than to durability.
Modern powertrains also produce documented, expensive failures. NHTSA imposed consent orders totalling US$210 million on Hyundai and Kia in November 2020 over untimely recalls of more than 1.6 million vehicles fitted with 2.0L and 2.4L Theta II petrol direct injection engines - a modern, high-output engine family, not an old design. Closer to home, the Federal Court declared by consent in April 2018 that Ford Motor Company of Australia engaged in unconscionable conduct in the way it dealt with complaints about Fiesta, Focus and EcoSport PowerShift dual-clutch transmissions between 1 May 2015 and 29 February 2016, and ordered $10 million in penalties. That declaration is about complaint handling rather than a contested finding on the transmission, but the ACCC records that over 70,000 vehicles were affected and that 37% of them required at least one clutch replacement.
The largest hole is the one nobody can fill. The question people are actually asking - how often did 1980s and 1990s cars need major work at 200,000 to 300,000 km - has no public dataset behind it. Mandatory public disclosure of product warranty liabilities began with FASB Interpretation No. 45, effective for financial statements of periods ending after 15 December 2002. There is no comparable earlier public series, and no ADAC or TÜV by-age equivalent reaching back that far on a consistent basis. Any confident number for that era is not coming from data.
Australia does not measure this
The most important thing to say to an Australian audience is that we have no national by-age vehicle defect dataset. Roadworthiness here is set by each state and territory rather than nationally, and the results are not published as a national by-age series. Every Australian argument about whether old cars were more reliable is currently being run on imported German, British and American numbers.
The British data is thinner than it looks. DVSA publishes MOT testing results for Great Britain broken down by vehicle class and defect category - not by vehicle age - and states the data set is not classed as an official statistic. Age-banded MOT failure rates circulate widely; they come from separate re-analyses that need naming, dating and linking, with their age bands and vehicle class stated, before anyone relies on them. They also carry the trap this whole article is about: cars still being presented for test at 25-plus years are a self-selected, better-maintained population, so an age gradient cannot by itself show that older cars are less reliable.
We could not point to a current, linkable national figure for the average age of the Australian fleet either. That gap matters, because the German inspection data above shows defect rates climbing out of single digits during a car's second decade, and there is no Australian series to check that against.
Age is not automatically safer, either. Takata's compulsory recall, issued in February 2018, covered more than 3 million Australian vehicles and over 4 million inflators, part of a global recall of an estimated 100 million vehicles. The propellant degrades with prolonged heat and humidity exposure, so the hazard variable is the age of the inflator and the climate it has lived in - materially worse in Queensland, the NT and northern WA. Globally over 350 injuries and 32 deaths were reported; in Australia, 1 death and 3 injuries including 1 serious injury. Manufacturers have completed 99.9% of the recall.
Australia's compensating advantage is legal rather than statistical. The acceptable-quality guarantee under the Australian Consumer Law has no fixed expiry, which is why Australia produces court-established facts about modern-car defects. In Williams v Toyota Motor Corporation Australia Ltd [2024] HCA 38, decided 6 November 2024, the High Court held that the reduction-in-value entitlement under s 272(1)(a) passes with ownership of the vehicle, while remitting the damages assessment to the trial judge to be redone taking account of the 2020 Field Fix. No final damages figure has been determined.
The corollary of damages passing with ownership is that a group member who has already sold the affected vehicle loses the reduction-in-value entitlement, keeping only any misleading-conduct and consequential-loss claims.
What this leaves you with
On the evidence we could reach, modern cars of a given age break down less often than cars of that same age did a decade ago, the regulatory durability floor has risen rather than fallen, and the era people remember as indestructible produced a fleet where fewer than one in ten cars reached twenty years registered. What has clearly got worse is the cost of standing behind a car, the number of problems owners report, and the consequences when a specified system - an after-treatment device, an inflator propellant, a dual-clutch transmission - is operated outside what it was specified for.
None of that tells you about the individual car in front of you. Age is the strongest single predictor in every compulsory-inspection dataset here, and none of those datasets is Australian. Inspect the car.