By on November 23, 2009

Maybe we should have checked a reliability survey... (

When buying a car, what do people really want to know about its reliability? Often: what are the odds it will turn out to be a lemon? And does it have a good shot at requiring no repairs at all?

Consumer Reports and J.D. Power have never answered these questions. They’ve only provided vague dot ratings that indicate how a car compares to the average for all cars. Even TrueDelta, which has been providing car models’ average repair frequencies, and not just dots, has not been directly answering these questions. Instead, car buyers have had to infer their odds of getting a lemon from the average repair frequency.

With the latest update to TrueDelta’s Car Reliability Survey results, released today, this will no longer be necessary. This update includes two new statistics: “Nada-odds” and “Lemon-odds.” From a car model’s Nada-odds, car buyers can learn how many cars out of a hundred required no repairs at all—nada—in the past year. And from its Lemon-odds they can learn how many out of a hundred had to go to the repair shop three or more times in the past year.

It turns out that the odds of getting a problem-free car are higher than most people probably suspect, while the odds of getting a lemon are probably much lower. Many of the 2007, 2008, and 2009 models included in these results had a three-in-four chance—or better—of requiring no repairs at all, not even a minor one, in the past year. With all but the least reliable fairly new cars the odds are at least 50-50. This is why, whenever a car model gets a “worse than average” reliability rating, there are plenty of owners who say this cannot be true because their car has required no repairs at all since they bought it a year ago.

And the Lemon-odds? With many fairly new cars the chances of a lemon are under one in a hundred, and under one-in-twenty is the norm. Only with the very least reliable cars are the odds worse than one-in-ten. The horror stories are real, just a lot less common than many people think.
These new stats require more data on more cars. So this initial set of results covers only 100 models, 64 of which are visible to the general public. As the number of participants grows, TrueDelta will provide these stats for more and more models.

With prompt quarterly updates, TrueDelta can provide reliability stats for new models well ahead of other sources. Three months ago TrueDelta released the first reliability stat anywhere for a 2010 model. The updated results, covering owner experiences through September 30, 2009, include additional 2010 models. The redesigned Toyota Prius and Mazda3 have required very few repairs so far, while the new Hyundai Genesis Coupe has been about average.

Car Reliability Survey results

Nada-odds and Lemon-odds

[Michael Karesh owns and operates TrueDelta]

Get the latest TTAC e-Newsletter!

11 Comments on “TrueDelta Announces Reliability Survey Results...”

  • avatar

    So my former 05 Odyssey lemon was that >1 in a 100.  Unlucky me.

  • avatar

    Thanks Michael.
    This is another interesting tool. 
    I guess I’m not at all surprised by the results on the Nada/Lemon odds.  Actually, I think a 10% chance of a “lemon” is pretty nasty and well worth avoiding.

    BTW, the usual suspects seem to be turning in the most consistent results on top reliability, how soon before the “defenders” of the weaker performers decide they have to accuse you of being corrupt and biased also.  Chuckle.

    Keep up the good work.


  • avatar

    Thanks Michael. Is there any way you could break out the VW ratings between the TDI and non TDI versions … assuming there’s a difference!

  • avatar

    The TDIs are broken out for the MkIV and the 2009. Any differences are usually due to the higher average odometer readings on the TDIs–they tend to be driven a lot.

  • avatar
    Brendon from Canada

    Hmmm…. the odds of getting a lemon actually sound way to high to me!   With a poor car being a 1 in 10 of being a lemon,  you would think that they’d be easy to spot, unless the vehicle was super-low volume (ie, our LR Freelander was in this group; good for us while we owned it, but the horror stories were more the norm, even from the dealership).    

    • 0 avatar

      Personally I don’t think three repairs in a year necessarily counts as a “lemon” — a high repair rate, yes, but if it’s three minor repairs, that wouldn’t make the car a lemon in my book.
      Good data, though.  While it has its limitations, it’s certainly no worse (and likely better) than other reliability surveys out there.

  • avatar
    Greg Locock

    Interesting stuff.  I wonder what proportion of newly built houses are NADAs and lemons, by these definitions?

  • avatar
    Jeff Puthuff

    Interesting! I’m quite surprised by the Sx4’s 96/<1 rating. Will have to keep an eye on Suzuki; hope they last in the US market.

  • avatar

    It’s really a shame given the high quality and value of Michael Karesh’s work, and that he only needs 25 cars of a given model and year, that so many models have inadequate rates of participation.   He has put a huge effort into spreading the word about truedelta, and has put together a very nicely functioning and useful piece of Internet value.

    How can it be that out of so many millions of car owners, many models and model years can’t get a 25-vehicle participation rate.  You’d think that even if a healthy proportion of TTAC members participated, the truedelta info would be far more encompassing.  I’m a contributoar to truedelta, and it really is minimal bother.

  • avatar

    I contribute as well, and it’s pretty easy – especially since my ’08 Elantra hasn’t needed any repairs (but it’s a 5spd with 6500 miles on it  ;-)

  • avatar

    Michael, I’d love to join TrueDelta, but I don’t give out my email address to any sites anymore.  Instead, I use with an extensive whitelist (which TD would immediately join).  Unfortunately, you’re still blocking spamgourmet addresses at the door – I’ve been checking back for years.
    I don’t think TD will be the source of any spam.  I just route all mail there so I can 1) track where I gave out the address and 2) not have to update 500 websites when I change email addresses.  I’d do this on my own domain if I ever got around to having the time, and you’d have no way to know that I’d given you a “throw-away” address.
    If the issue is getting surveys filled out, why not track survey completions instead?

Read all comments

Back to TopLeave a Reply

You must be logged in to post a comment.

Recent Comments

  • FreedMike: This youtube video proves everything you ever wanted to know about everything. Just watch it....
  • Imagefont: Well you made me look. Now I want the ‘69 white Alpha Spider with the long boat tail – beautiful...
  • jack4x: “I tend to troll for listings for my cars.” I also do this, so I keep an eye on 5th gen Vipers,...
  • Art Vandelay: B13 Sentra SE-R (and NX2000) Z32 300Z E39 540i (I was second on a low mileage one and the winner backed...
  • Old_WRX: I did find the video after all. v=Du2wm5nhTXY Have you seen this? Comments?

New Car Research

Get a Free Dealer Quote

Who We Are

  • Adam Tonge
  • Bozi Tatarevic
  • Corey Lewis
  • Mark Baruth
  • Ronnie Schreiber