The 2017 model year, ranked by risk
The 2017 field, worst and best
Of the 263 vehicles from the 2017 model year carrying enough owner reports to score, the Hyundai Santa FE comes out worst at 85 out of 100 and the Buick Regal best at 58. Every score on this page is ranked against 2017 cars only.
What a rank inside one model year means
Older cars have had more years to break and more years to accumulate complaints, so comparing a 2017 against a car ten years newer measures age as much as quality. Scoring inside a single model year removes that: every vehicle on this page has had the same 9 years on the road.
It does not correct for popularity on its own, either. The rate per 1,000 vehicles does that, using fleet sizes derived from recall records; where no recall exists to derive one from, the vehicle is scored on counts and marked as lower confidence.
Highest risk
The worst-scoring vehicles of 2017 that also have a fleet estimate, so a rate can be worked out. Shorter is better. Where the top bar towers over the rest, the model year has one outlier rather than a bad field; where the bars are level, the whole cohort was ordinary and the ranking below is splitting hairs.
What each column means, and what it does not
Rank is the vehicle's position among the 263 vehicles of 2017 that carry enough owner reports to be scored at all. A car missing from this page is not a clean car; it is a car too few people filed about, which for a rare model is the normal outcome. Rank is the whole point of publishing the cohort: a vehicle described elsewhere on this site as the seventh riskiest of its year can be checked against the other six here.
Reports is the raw count of owner complaints filed with NHTSA about that vehicle, and it is the column most likely to mislead you. A model that sold two million copies will out-report a model that sold forty thousand without being any worse to own, which is why the column beside it exists. Per 1,000 divides those reports by an estimate of how many of that model year are on the road, derived from the fleet sizes manufacturers state in their recall filings. It is an estimate and it is blank wherever no recall covers the model year — roughly the cleanest vehicles and the rarest ones, an absence that cuts both ways.
Median report is the mileage by which half the dated reports had been filed, and it is the number a buyer standing in a driveway actually needs. Two cars with identical complaint counts are not the same car if one reports at 40,000 miles and the other at 140,000: the first is a defect, the second is wear. Most reported problem names the failure mode with the highest weighted report count for that vehicle, and Risk is the 0–100 score that combines rate, severity and timing, ranked inside 2017 and nothing else.
Nothing on this page is a measurement of how long a car lasts. It is a measurement of what owners chose to report to a federal complaints database, which is a real signal about defects and a poor one about oil changes. Read it as a list of things to have inspected, not as a verdict.
How far apart these cars really are
Across the 40 ranked vehicles shown, scores run from 85 down to 80. The distance between neighbours in the middle of a list like this is often a point or two, which is well inside the noise of voluntary reporting — read the ends of the ranking, not the middle of it.
Taking the worst failure mode from each of these 2017 vehicles, the problems that recur are other engine problem (13 vehicles), other powertrain problem (6 vehicles) and airbag failed to deploy in a crash (4 vehicles). A fault that shows up across several manufacturers in the same model year is usually a component they all bought from the same supplier.
The 2017 cars that hold up best
Lowest scores in the 2017 field: Buick Regal (58), Cadillac Xts (59), Ford Transit Connect (59), Lexus NX (60) and Tesla Model 3 (60).
A low score means owners of that car reported fewer serious failures, relative to how many were built, than owners of its contemporaries. It says nothing about the car in a particular driveway, and it is not a substitute for an inspection or a service history.
Other model years
- 2026 model year
- 2025 model year
- 2024 model year
- 2023 model year
- 2022 model year
- 2021 model year
- 2020 model year
- 2019 model year
- 2018 model year
- 2017 model year
- 2016 model year
- 2015 model year
- 2014 model year
- 2013 model year
- 2012 model year
- 2011 model year
- 2010 model year
- 2009 model year
- 2008 model year
- 2007 model year
- 2006 model year
- 2005 model year
- 2004 model year
- 2003 model year
- 2002 model year
- 2001 model year
- 2000 model year
- 1999 model year
- 1998 model year
- 1997 model year
- 1996 model year
Using this list to shop
A ranking of every 2017 model year is most useful in reverse: it tells you which cars of that age have a documented pattern of expensive failures, so you know what to look for and what to walk away from. The model-year page for each car names the failures and the mileages they are reported at.
Scores are recomputed whenever the complaint database is re-imported, and a car's position can move as reports accumulate. The figure shown is the one from the most recent computation, and every page carries its own date.
Questions this page answers
Which 2017 vehicles are the least reliable?
Of the 263 2017 vehicles with enough owner reports to score, the Hyundai Santa FE carries the highest computed risk at 85 out of 100. Scores are ranked within the 2017 model year only, so a car is never penalised for being older.
Which 2017 vehicles hold up best?
The Buick Regal scores lowest at 58 out of 100 among 2017 vehicles with enough reports to rate. A low score means few reported failures relative to how many were built, not a guarantee.
What a low score is not. It is not a promise. A vehicle can sit at the safe end of its cohort because it genuinely had fewer reported defects, or because the people who bought it complain less, or because fewer were sold than our fleet estimate assumes. The score is a comparison of complaint behaviour, weighted by how bad each failure is, and the methodology page is explicit about which parts of that we are confident in.
Source Owner complaints and recall records published by the US National Highway Traffic Safety Administration, model years 1996–2027 Figures rebuilt Published by AtWhatMileage How every figure is computed