Volkswagen Golf years to avoid: the early ones
Based on 873,808 real MOT tests from DVSA records. Each year is compared against every other car first registered the same year — otherwise the numbers would only tell you that older cars fail more, which you knew.
The short answer
Avoid 1998. The worst is the 1998, which fails its MOT 16% more often than other cars of the same age. The 2015 is the strongest, failing 34% less often than average.
Found a Volkswagen Golf you like?
Averages tell you what to expect. Its registration tells you what actually happened — every MOT, every advisory, and whether the mileage adds up.
Every Golf model year
Shaded rows are the years that differ most from other cars of the same age — worse, better.
| Registered | MOT tests | Fail rate | vs same-age cars |
|---|---|---|---|
| 1990 | 1,704 | 24.2% |
about average
|
| 1991 | 1,737 | 24.6% |
about average
|
| 1996 | 1,024 | 30.1% |
about average
|
| 1997 | 1,075 | 31.0% |
about average
|
| 1998 Worst | 1,505 | 36.1% |
16% worse
|
| 1999 | 3,325 | 36.5% |
about average
|
| 2000 | 5,585 | 36.5% |
about average
|
| 2001 | 9,694 | 36.0% |
about average
|
| 2002 | 16,886 | 35.9% |
about average
|
| 2003 | 19,728 | 34.9% |
about average
|
| 2004 | 16,218 | 40.6% |
about average
|
| 2005 | 24,869 | 38.5% |
about average
|
| 2006 | 34,857 | 38.3% |
about average
|
| 2007 | 46,123 | 35.5% |
about average
|
| 2008 | 45,112 | 33.9% |
about average
|
| 2009 | 47,389 | 30.6% |
about average
|
| 2010 | 47,738 | 31.1% |
about average
|
| 2011 | 52,407 | 28.7% |
about average
|
| 2012 | 53,159 | 25.0% |
about average
|
| 2013 | 55,784 | 18.5% |
28% better
|
| 2014 | 65,163 | 15.9% |
30% better
|
| 2015 | 68,107 | 12.9% |
34% better
|
| 2016 | 65,746 | 11.3% |
33% better
|
| 2017 | 70,372 | 10.4% |
28% better
|
| 2018 | 62,380 | 9.2% |
23% better
|
| 2019 | 49,095 | 8.7% |
15% better
|
| 2020 | 7,026 | 6.9% |
25% better
|
How this is worked out
Every MOT test recorded by DVSA in 2023 on a Volkswagen Golf, grouped by the year the car was first registered. Retests are excluded, so a car that failed and was fixed still counts as a failure. Each year's rate is then divided by the rate for all cars first registered that year — so 120 means 20% more failures than its contemporaries, not 20% more than a brand new car.
Contains public sector information licensed under the Open Government Licence v3.0.