Mazda Cx-3 years to avoid — 73,515 MOT tests ranked
Every Cx-3 year, ranked
The age-adjusted MOT index sets 100 as the pass rate of all UK cars registered in the same year and tested in the same calendar year. Higher is better. This prevents an old car being labelled a “bad year” simply because it is old.
| Rank | Year | Index | Vs same-age | Pass rate | Tests |
|---|---|---|---|---|---|
| 1 | 2018 Strong | 104.8 | +4.8 pts | 91.1% | 16,668 |
| 2 | 2017 Strong | 104.3 | +4.3 pts | 88.3% | 14,373 |
| 3 | 2019 | 104.2 | +4.2 pts | 92.6% | 20,957 |
| 4 | 2015 | 102.2 | +2.2 pts | 80.9% | 6,996 |
| 5 | 2016 | 100.8 | +0.8 pts | 82.1% | 14,521 |
How to use this before buying
- Rankings describe the cohort, not the car. A cherished older example beats a neglected 2017 one — always run the specific reg through the free MOT history check.
- Read the advisories. Repeated notes for the same component across years mean the owner has been ignoring it. Decode them in our advisory library.
- Scan before you pay. A 2-minute OBD scan reveals stored faults the dashboard hides — see our scanner guide.
FAQs
Which Cx-3 years should I avoid?
No year is a clear negative outlier after correcting for age, so buy on condition and history rather than year.
What is the best Cx-3 year to buy used?
2017 and 2018 lead the age-adjusted MOT ranking. The score compares each registration year with all UK cars of the same age; the specific car's history and condition still take priority.
Is the Cx-3 reliable overall?
Across 73,515 tests, the median age-adjusted index is 104.2 (100 equals the pass rate of all cars registered in the same year). Use the spread between years as one signal, then check the individual registration.
Methodology: index = 100 + the model-year cohort’s MOT pass-rate difference versus all UK vehicles registered in the same year and tested in the same calendar year. Class 4 initial tests only; retests excluded; at least 1,500 tests per page. “Inspect” years are below both the same-age baseline and the model’s median index by at least 2 points. Association is not proof of a model-year defect. DVSA anonymised data 2023–2025, OGL v3.0.