Used cars offered for sale in another country far more often carry a stated mileage below their last official odometer reading
3.90% of cars offered for sale in another country (301 of 7,715) vs 0.53% of cars offered for sale at home (465 of 87,525) had a stated mileage below their last official odometer reading
- Last updated
- Reading time
- 13 min
- Methodology
- v1.0
Key takeaways
- Cars offered for sale in a different country from the one where they were last inspected were offered below their last official odometer reading in 3.90% of cases (301 of 7,715; 95% CI 3.49–4.36%): about one car in 26.
- Cars offered for sale in the country of their last inspection: 0.53% (465 of 87,525), about one in 190, no higher than the rate of obvious data errors in the other direction.
- After adjusting for seller type, mileage, time since the inspection and which country's records were used, cars offered in another country were about four times as likely to be offered below their official reading (odds ratio 4.1; 95% CI 3.3–5.1).
- The gap is largest for high-mileage cars: 11.84% (225 of 1,901) of cars offered abroad whose last official reading was 200,000 km or more, against 1.16% (260 of 22,482) at home.
- Across all cars offered for sale in Poland with an earlier Danish, Czech or Slovenian reading: 4.29% (189 of 4,409), about one in 23.
- Czech cars sold in the Czech Republic show no detectable excess: 0.32% (205 of 64,158), below the 0.49% background.
- A discrepancy is a measurable warning sign for a buyer, not proof of fraud: a replaced instrument cluster, a typing error by the seller or a record error can explain one.
Carlytics compared the mileage stated by sellers of used cars across Europe with the odometer readings that government inspectors had already recorded for the same car, matched on the full 17-character VIN. Among cars offered for sale in the same country where they were last inspected, a stated mileage clearly below the official record was rare: 0.53% (465 of 87,525 cars), no higher than the rate of obvious data errors we see in the other direction. Among cars offered for sale in a different country from the one where they were last inspected, it was 3.90% (301 of 7,715; 95% CI 3.49–4.36%): about one car in 26 offered in another country, against one in 190 at home. After adjusting for seller type, mileage, time since the inspection and which country's records were used, cars offered in another country were about four times as likely to be offered below their official reading (odds ratio 4.1; 95% CI 3.3–5.1).
What we measured
An odometer only moves forward. If an inspector recorded 250,000 km and the car is later offered for sale with 180,000 km, one of these is true: the odometer was wound back, the instrument cluster was replaced, the seller's stated mileage is wrong, the inspection record is wrong, or the VIN given by the seller belongs to another car. We call this an odometer discrepancy. It is a measurable warning sign for a buyer, not proof of fraud in any individual case, and this study makes no claim about any car, seller or dealer.
- Used-car data: Carlytics' records of used cars offered for sale in 24 European countries, of which 1.2 million cars carried a valid VIN and a seller-stated mileage between April and September 2026.
- Official readings: public inspection records that include the odometer and the full VIN: the Czech Ministry of Transport's open data on periodic technical inspections (2008–2026), the Danish Motor Register (latest inspection per vehicle; mileage published in whole thousands of km, so we used the lower bound), and Slovenia's open roadworthiness-test data (to January 2023).
- Matched cars: 95,240 cars offered for sale had at least one official reading dated before the car was first offered.
- Discrepancy: the seller states at least 5,000 km, and at least 5%, less than the most recent official reading before the car was first offered.
Results
Cars offered for sale at home vs cars offered in another country
Share of cars offered with a stated mileage at least 5,000 km and 5% below their last official odometer reading
- Offered in another country
- Offered in the country of the last inspection
All matched cars
Czech inspection records only
Danish inspection records only
Private sellers
Dealers
Last official reading ≥ 200,000 km
Last official reading < 200,000 km
| group | cars | offered below official reading | 95% CI | "impossible increase" background† |
|---|---|---|---|---|
| Offered in the country of the last inspection | 87,525 | 0.53% (465) | 0.49–0.58% | 0.52% (451) |
| Offered in another country | 7,715 | 3.90% (301) | 3.49–4.36% | 0.58% (45) |
† The background is the share of cars whose stated mileage shows an implausibly large increase over the official reading (more than 20,000 km and more than 200 km a day). It measures how often a seller's stated mileage and a record plainly do not describe the same odometer: mismatched VINs, typing errors. Because this background requires more than 200 km a day, it shrinks for older readings; the at-home comparison holds under a time-independent check as well (cars at home with a reading from the last two years: 0.55% offered below the reading, against 0.99% offered more than 50%, and more than 50,000 km, above it). At home, the discrepancy rate is at the level of this background. For cars offered in another country it is 3.90%: 7.3 times the at-home rate before adjustment; comparing cars checked against the same register, 3.7 times (Danish records) and 8.7 times (Czech records).
The gap holds within every slice we checked (post-hoc robustness checks, see Methodology):
| slice | offered in another country | offered at home |
|---|---|---|
| Czech inspection records only | 2.77% (45 / 1,625) | 0.32% (205 / 64,158) |
| Danish inspection records only | 4.10% (248 / 6,050) | 1.11% (260 / 23,362) |
| Private sellers | 4.92% (240 / 4,878) | 0.89% (170 / 19,204) |
| Dealers | 1.91% (51 / 2,677) | 0.45% (272 / 60,573) |
| Last official reading ≥ 200,000 km | 11.84% (225 / 1,901) | 1.16% (260 / 22,482) |
| Last official reading < 200,000 km | 1.31% (76 / 5,814) | 0.32% (205 / 65,043) |
Where the cars were offered for sale
Exploratory; not corrected for multiple comparisons.
| offered in | last official reading in | cars | offered below official reading | 95% CI |
|---|---|---|---|---|
| Poland | Denmark | 3,766 | 4.20% (158) | 3.60–4.88% |
| Poland | Czech Republic | 620 | 4.03% (25) | 2.75–5.88% |
| Romania | Denmark | 498 | 9.84% (49) | 7.52–12.77% |
| Germany | Denmark | 650 | 2.15% (14) | 1.29–3.58% |
| Czech Republic | Czech Republic | 64,158 | 0.32% (205) | 0.28–0.37% |
| Denmark | Denmark | 23,362 | 1.11% (260) | 0.99–1.26% |
Country pairs with fewer than 100 matched cars or fewer than 10 discrepancies are not shown.
- Across all cars offered for sale in Poland with an earlier Danish, Czech or Slovenian reading: 4.29% (189 of 4,409; 95% CI 3.73–4.93%), about one in 23. 88.8% of these cars were offered by private sellers.
- In Romania: 9.35% (56 of 599; 95% CI 7.27–11.95%), about one in 11 (between one in 8 and one in 14). This rests on about 600 cars, most of them (498) with a Danish record.
- Where the stated mileage was below the record, the typical shortfall for cars offered in another country was 54,000 km (middle half: 25,000–103,000 km), about 22% of the last official reading.
Other patterns
- High-mileage cars: 1.33% of cars with a last official reading of 150,000 km or more were offered below it, against 0.35% below 150,000 km (pre-registered comparison; difference 0.98 points, 95% CI 0.86–1.10).
- Private sellers vs dealers: 1.70% of private-seller offers vs 0.51% of dealer offers (pre-registered; difference 1.19 points, 95% CI 1.03–1.37). Dealer offers were at the level of the background measure (0.52%).
- Czech cars sold in the Czech Republic: cars offered there with a Czech inspection history were offered below their official reading in 0.32% of cases, below the 0.49% background, and also below the time-independent check (cars with a reading from the last two years: 0.33% vs 1.07%). There is no detectable excess. Since January 2015 the Czech Ministry of Transport has let anyone look up a car's inspection odometer readings online (kontrolatachometru.cz); today the service needs only the VIN.[1] This study cannot say whether that is the reason.
What this does and does not mean
- It means: a car offered for sale in a different country from the one where it was last inspected is several times more likely to be offered with fewer kilometres than an official inspector recorded, and the gap is largest for high-mileage cars offered by private sellers.
- It does not mean: that 4% of cars in Poland or 10% of cars in Romania have tampered odometers. We can only see cars whose previous inspection records are public and VIN-linked (mostly cars previously inspected in Denmark or the Czech Republic). Cars imported from countries that do not publish VIN-linked inspection readings, including Germany, are invisible here, and a reduction made before the last official inspection cannot be detected at all. The true share of wound-back odometers among imports could be higher or lower.
- It does not identify anyone. Individual discrepancies have innocent explanations (a replaced instrument cluster, a typing error by the seller). The buyer's takeaway is to check a car's official odometer history against the mileage the seller states before paying.
Limitations
- Who is in the sample. Only cars in Carlytics' used-car data with a VIN given by the seller and with a public VIN-linked inspection record. Cars offered for sale in the Czech Republic dominate the matched set (68%); Poland and Romania enter only through cars previously inspected in Denmark and the Czech Republic. A seller who has lowered an odometer has a reason not to show the VIN, so offers that do show it probably understate the rate.
- Data errors. A hand review of 40 randomly chosen discrepancies found 31 consistent with a reduced or replaced odometer and 9 (22.5%; 95% CI 12–38%) explained by record errors or errors in the seller's stated mileage, or left unclear, mostly same-country Danish cases (digits dropped in a stated mileage; Danish register values mis-scaled on newly registered cars). Among the 18 cars in that sample offered in another country, 16 had no sign of a data error. The at-home rates above therefore overstate true reductions; the difference between cars offered abroad and at home holds within each register separately (Danish: 4.1% vs 1.1%; Czech: 2.8% vs 0.3%).
- One Danish reading per car. The Danish register publishes only the latest inspection, in whole thousands of km. We used the lower bound of each band, which makes a discrepancy less likely, not more.
- Offer dates. A stated mileage is a claim made when the car was first offered. We compared only with official readings dated before the car was first offered. One archive-like source whose records turned out to show years-old sales was removed after the first run (see Methodology).
- Identity. Offers and records were matched on the full VIN, and offers whose make disagreed with the register were excluded. Among the discrepancy cars offered in another country, the seller's stated model agreed with the register in 92.6% of checked cases (262 of 283; 92.1% for the other cars offered abroad; most disagreements are naming, such as "Klasa E" against "E350") and the year was within one year of first registration in 97.9% (282 of 288; 99.5% for the other cars offered abroad). Very few of these cars were also offered for sale in the country of the record, so identity rests on these checks rather than on an independent second sighting.
- Thresholds. The headline uses a 5,000 km / 5% tolerance. The gap between cars offered abroad and at home holds at every tolerance tested:
| tolerance | offered in another country | offered at home | ratio |
|---|---|---|---|
| > 1,000 km | 6.56% (506) | 1.26% (1,103) | 5.2× |
| > 5,000 km and 5% (headline) | 3.90% (301) | 0.53% (465) | 7.3× |
| > 10,000 km and 10% | 3.18% (245) | 0.37% (327) | 8.5× |
| > 25,000 km and 20% | 2.03% (157) | 0.28% (242) | 7.4× |
(denominators: 7,715 offered in another country, 87,525 at home)
- Timing. The cars were offered for sale between April and September 2026; official readings span 2008–2026 (median June 2025). Our copy of the Czech records is complete to February 2026; March 2026 holds about two-thirds of a normal month and April–August 2026 only fragments. Missing recent records can only make the reference reading older and lower, again the conservative direction.
Methodology
v1.0Population. Observations of used cars offered for sale with a valid 17-character VIN, a mileage between 1,000 and 1,500,000 km, from 24 European countries (26 were eligible; Lithuania and Luxembourg contributed no eligible offers; UK and Ireland excluded for miles), 1 April – 22 September 2026, excluding archive datasets. One offer per car (latest observation; among same-day offers with the same mileage, the one with the earliest offer date). 3,435,092 observations → 2,888,419 eligible → 1,226,746 cars → 103,124 with an eligible official record → 7,749 with no reading before the offer → 135 excluded because the offer and the register named different makes → 95,240 cars.
Reference reading. The most recent official odometer reading dated before the car was first offered (earliest of: the stated offer date, the date first recorded and the observation date). Czech and Slovenian records restricted to passenger cars (category M1); Danish records with an active inspection status, lower bound of the published thousand-km band.
Outcomes. Discrepancy: official − stated > max(5,000 km, 5% of official). Background: stated − official > 20,000 km and > 200 km per day since the reading.
Statistics. Wilson 95% intervals for proportions; Newcombe intervals for differences; two-proportion tests with Holm correction across the three pre-registered comparisons (high mileage, private vs dealer, offered abroad vs at home; all p < 10⁻⁶ after correction); Mantel–Haenszel odds ratio across strata of seller type × mileage band × time since reading × register, with a stratified bootstrap CI (500 resamples): OR 4.07 (3.27–5.08; analytic Robins–Breslow–Greenland CI 3.36–4.93), risk difference 2.75 points (2.27–3.25). Cells under 100 cars or with fewer than 10 discrepancies are not reported.
Sensitivity. Adjusted odds ratios across all specifications tested (adding an age band; excluding implausible official readings; excluding dropped-digit offers; excluding cars still registered in Denmark; and combinations of these) ranged from 3.8 to 5.0. Excluding 61 cars offered abroad that were still registered in Denmark (possible identity mismatch): 3.68% (282 of 7,654); adjusted OR 3.8 (3.1–4.6). About 98% of the Danish-record cars offered abroad were deregistered in Denmark. The contrast is not concentrated in offers made shortly after the inspection: for cars offered abroad the rate was 3.3% where the last reading was under a year old and 4.6–4.8% where it was more than two years old.
Pre-registration and deviations. The question, outcome definitions, thresholds and the three comparisons were written down before any matching was run. Four disclosed amendments: (A1) tie-breaks and make aliases, before results; (A2) after the first run, the offer date was changed from "date observed" to "date first appeared" and one archive-like source was removed, because its records showed sales from 2021–2025; this lowered the overall rate from 1.61% to 0.81%; (A3) the stratified and sliced checks above were added after seeing results; (A4) after independent review, a tie-break for two same-day offers of one car with the same mileage (take the earliest offer date) and three more make aliases (Mercedes-AMG, Abarth, DS Automobiles); no headline figure moved by more than 0.04 points. Leading with the contrast between cars offered abroad and at home (pre-registered as hypothesis H4) rather than the overall rate was decided after seeing the results.
Verification. Every headline figure was computed twice before the final tie-break rule: once from extracted files, once directly in the database from a different table with independently written logic. They agreed within 0.05 percentage points (e.g. discrepancy rate 0.806% vs 0.818%; cars offered abroad 3.91% vs 3.86%). A third, independently written computation agreed within 0.1 points. An independent reviewer then recomputed every headline cell from the raw extracts, matched the first computation exactly, and applied the final tie-break rule; the figures in this article are from that final run (overall 0.80%, 766 of 95,240). A 40-case hand review was also run.
Not reported. A planned follow-up (cars offered for sale in Germany that later received an official reading abroad) matched only 250 cars, below our pre-set minimum of 300 for a headline, and is not reported.
Privacy. Only aggregates are published. No VIN, registration plate, individual offer, seller or location below country level appears in this study or its tables.
Frequently asked questions
›Does this mean 4% of cars in Poland have a tampered odometer?
No. We can only see cars whose previous inspection records are public and VIN-linked (mostly cars previously inspected in Denmark or the Czech Republic). Cars imported from countries that do not publish VIN-linked inspection readings, including Germany, are invisible here, and a reduction made before the last official inspection cannot be detected at all. The true share of wound-back odometers among imports could be higher or lower.
›Is a stated mileage below the official reading proof of odometer fraud?
No. If an inspector recorded 250,000 km and the car is later offered for sale with 180,000 km, one of these is true: the odometer was wound back, the instrument cluster was replaced, the seller's stated mileage is wrong, the inspection record is wrong, or the VIN given by the seller belongs to another car. It is a measurable warning sign for a buyer, not proof of fraud in any individual case, and this study makes no claim about any car, seller or dealer.
›Which official odometer records does the study use?
Public inspection records that include the odometer and the full VIN: the Czech Ministry of Transport's open data on periodic technical inspections (2008–2026), the Danish Motor Register (latest inspection per vehicle; mileage published in whole thousands of km, so we used the lower bound), and Slovenia's open roadworthiness-test data (to January 2023).
›What should a buyer do with this?
Check a car's official odometer history against the mileage the seller states before paying. Since January 2015 the Czech Ministry of Transport has let anyone look up a car's inspection odometer readings online (kontrolatachometru.cz); today the service needs only the VIN.
Sources
- [1]Press release on the Czech online odometer lookup (service live from 2 January 2015) - Czech Ministry of Transport (2015-01-05). https://md.gov.cz/Media/Media-a-tiskove-zpravy/MD-brani-podvodum-pripravilo-novou-aplikaci-na-ko
- [2]Report on the launch of the Czech inspection odometer lookup - IZDoprava (2015-01-07). https://izdoprava.cz/stav-z-tachometru-z-technicke-kontroly-je-mozne-vyhledat-na-internetu/
- [3]kontrolatachometru.cz: at launch the lookup needed the VIN and the inspection report number and showed the last inspection; the live service asks only for the VIN and shows the readings recorded at technical inspections and emission tests (checked 23 September 2026) - Czech Ministry of Transport. https://www.kontrolatachometru.cz/