Torque converter failure on the 2006 Acura MDX
When torque converter failure gets reported
Owners of the 2006 Acura MDX report torque converter failure at a median of 70,000 miles. A quarter of the reports come in before 55,000 miles and a quarter after 80,000 miles, from 15 reports that carried an odometer reading.
A part like this is normally expected to survive 96,000 miles. It is the 2nd most-reported problem on this model year.
The mileage distribution
The reported range runs from 50,000 miles at the tenth percentile to 90,000 miles at the ninetieth, a spread of 38,634 miles across the middle 80% of reports.
A window that narrow is the signature of a part that fails on its own schedule rather than on how the car is driven — which is usually a design or supplier problem rather than wear and tear.
0% of the dated reports were filed under 36,000 miles.
The whole distribution
| 10th | 25th | Median | 75th | 90th |
|---|---|---|---|---|
| 49,966 miles | 57,000 miles | 68,500 miles | 78,000 miles | 88,600 miles |
Read the 10th percentile as “one report in ten arrived this early”. Exact figures are shown in this table because it is the table of record; everywhere else on the site they are rounded to the precision the sample can carry.
What a percentile can and cannot tell you here
These five numbers describe the readings owners gave, in order. The median is the middle one: half the reports arrived below it. The 10th and 90th are the edges of the bulk of the distribution, and the distance between them is the part worth reading. A tight spread — say 70,000 to 95,000 miles — points at something that wears out on a schedule, and the schedule is knowable. A wide one, 20,000 to 190,000, points at a fault that hits some cars and never troubles others, which means the mileage on the listing in front of you tells you very little about your odds.
What these figures are not is a survival curve. Every reading here comes from a car whose owner filed a complaint, so the sample contains only failures: it cannot say what share of the fleet made it past 150,000 miles untouched, because those owners had nothing to report. Nor is the mileage the moment of failure — it is the odometer reading as recalled at the time of filing, which may be weeks later and is sometimes rounded to the nearest ten thousand by the person typing it.
Percentiles are shown at all only once 8 readings exist for the vehicle and mode. Below that the middle of a handful of numbers is noise wearing a decimal point, so the figures are withheld rather than dressed up.
What happens, and what it costs
Putting it right typically runs $1,100–$3,200 at an independent US shop. That range is an editorial estimate held in this site's failure-mode catalogue, not a figure from NHTSA, and it is a repair cost rather than a diagnosis cost.
The same failure on other years
The same failure is reported on 11 model years of this car. It shows up earliest on the 2009, at a median of 35,000 miles, and latest on the 2003 at 120,000 miles.
By volume the 2005 carries the most reports, with 38; the 2006 has 16. A mode that appears across most years of a model is usually a design choice rather than a bad batch.
If one year stands out to the left, that is the year to avoid; if the whole line sits at a similar mileage, this is wear rather than a defect in one production run.
Tick median reported mileage Solid middle half of reports Faint 10th to 90th percentile Colour how bad the failure is when it happens
What owners wrote
Verbatim from NHTSA complaint narratives, trimmed for length. The mileage is the one the owner gave.
2006 acura mdx with 68,500 miles has a shudder feeling at speeds between 35 and 45 MPH. The vehicle has been into the dealer twice and the service technician has stated that it is a torque converter problem, however refuse to fix the problem. *tr
Tl*the contact owns a 2006 acura mdx. When the contact drove the vehicle between 35-40 MPH, the contact heard a rumbling sound coming from the drive train and the vehicle began to vibrate. The vehicle was taken to the dealer where the contact had the torque converter and the ecu replaced. There were no prior warnings. The current and failure mileages were 67000.
I have a 2006 mdx and when it had about 55,000 miles on it I started to notice a shaking/lurching feeling when going about 30-40mph while accelerating. I took the car into and acura dealership and they replaced the torque converter, under warranty. When I picked the car up they said that it should never happen again and if it did then there should be some kind of recall done. Well, 10,000 miles later and it is happening again. I took the car back to the dealership and of course they can't recreate the problem which means they don't have to fix the real problem. The car is no longer under warranty!! *tr
Also reported on the 2006 Acura MDX
Other powertrain problem
Complete transmission failure
Seat belt or retractor failure
Other engine problem
Other brake problem
Airbag inflator rupture
How far this goes
Complaints are voluntary, so 16 is a floor and not a count of affected cars. Mileages are typed in by owners and cluster on round numbers. And the mode itself is assigned by software reading the complaint text, which is reliable in bulk and fallible one report at a time.
Questions this page answers
At what mileage does torque converter failure happen on the 2006 Acura MDX?
Across 15 owner reports that included an odometer reading, half were filed by 70,000 miles. A quarter were filed before 55,000 miles and a quarter after 80,000 miles. These are the mileages owners reported when they complained, not a prediction for any individual car.
How common is torque converter failure on the 2006 Acura MDX?
16 reports have been filed with NHTSA, about 0.1 per 1,000 vehicles. Complaints are voluntary, so this is a floor rather than a count of affected cars.
What this is and is not. These are the mileages at which owners chose to complain to a federal agency, which is not the same as the mileage at which the part fails on every car. It skews toward failures that happened outside warranty, cost real money, or frightened someone. It is still the only public dataset that carries failure mileage at all, and the shape of a few hundred readings is informative even when a single one is not. Methodology · JSON · this failure on other cars
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