Updated 14 August 2026. Shortly after first publication this study was revised to credit existing work on this topic by NHTSA, Carfax and others, and to state its sampling frame more precisely. The findings are unchanged; the context around them was incomplete. We restate corrections rather than editing silently.
Most recalls ask you to book an appointment. A small number tell you to stop driving, or to stop parking near your house. We analysed 60 of the 62 such United States campaigns we currently hold. The result is not what the headlines suggest: the typical urgent recall is tiny, and almost none of them can be fixed by software.
The population
| Measure | Value |
|---|---|
| Campaigns analysed | 60 (of 62 held) |
| Carrying do not drive | 27 |
| Carrying park outside | 34 |
| Carrying both | 1 |
| Report dates covered | 11 April 2025 to 30 July 2026 |
| Vehicles potentially affected, in total | 2,422,750 |
What was already known
This topic has occupants, and it would be dishonest to present the population itself as a discovery.
NHTSA publishes annual totals. Its 2025 Annual Recalls Report charts vehicles recalled under do-not-drive and park-outside advisories by recall year, going back to 2021. Carfax maintains a public park-outside tracker and reported in June 2026 that such recalls had risen by nearly half, covering more than 3.2 million vehicles. vehiclesafetyrecalls.com maintains live counts of active critical recalls, listing figures substantially larger than our holding.
What none of them publishes is the distribution. Totals tell you the scale of the category. They do not tell you that most campaigns in it are minuscule, that a handful carry nearly all the vehicles, or that almost none can be fixed remotely. That is what this study adds, and it is the whole of what it adds.
Finding 1: the typical urgent recall is very small
Most urgent recalls are invisible, and a handful carry the headlines. The mean campaign covers 40,379 vehicles; the median covers 629. A heavy-tailed distribution is not itself surprising: what matters is how extreme it is here, and what it means for how these recalls get reported.
36 of the 60 campaigns affect fewer than 1,000 vehicles. Twenty-one affect fewer than 100. The smallest in the set covers a single vehicle.
Meanwhile four campaigns account for 85% of all the vehicles in the population, and one accounts for 44% on its own: 26V363, Chrysler, a hydraulic power-assist steering pump, 1,076,999 vehicles, park outside.
So “millions of cars told not to park in a garage” and “most of these recalls involve a few dozen cars” are both true at once. Reporting that quotes only the total gets the shape of the risk wrong.
Each bar is one campaign, largest first. Four campaigns carry 85% of all 2,422,750 vehicles; the median campaign covers 629. Read 14 August 2026.
Show all 60 campaigns as a table
| Campaign | Manufacturer | Vehicles |
|---|---|---|
| 26V363 | Chrysler (FCA US, LLC) | 1,076,999 |
| 26V430 | Kia America | 462,869 |
| 25V741 | Chrysler (FCA US, LLC) | 320,068 |
| 25V636 | BMW of North America | 199,674 |
| 26V106 | Subaru of America | 69,870 |
| 25V390 | General Motors | 62,468 |
| 25V893 | Hyundai Motor America | 51,587 |
| 26V441 | BMW of North America | 29,119 |
| 25V798 | Kawasaki Motors Corp., U.S.A. | 19,823 |
| 25V376 | Kawasaki Motors Corp., U.S.A. | 17,792 |
| 26V362 | Chrysler (FCA US, LLC) | 17,277 |
| 25V503 | International Motors | 16,553 |
| 26V090 | Ford Motor | 15,965 |
| 25V414 | International Motors | 12,004 |
| 26V073 | Mercedes-Benz USA | 11,895 |
| 25V748 | BMW of North America | 5,361 |
| 25V502 | International Motors | 4,764 |
| 26V340 | Ford Motor | 4,653 |
| 25V628 | Ford Motor | 4,632 |
| 26V367 | Yamaha Motor Corporation, USA | 4,173 |
| 25V778 | Yamaha Motor Corporation, USA | 2,715 |
| 25V401 | Ford Motor | 2,345 |
| 26V257 | Winnebago Industries | 1,537 |
| 26V242 | Roush Performance Products | 1,217 |
| 26V197 | Yamaha Motor Corporation, USA | 962 |
| 26V028 | Volkswagen Group of America | 670 |
| 25V487 | Mercedes-Benz USA | 660 |
| 25V631 | New Flyer of America | 655 |
| 26V351 | Mack Trucks | 644 |
| 25V836 | Volkswagen Group of America | 629 |
| 25V291 | Hyundai Motor America | 620 |
| 25V238 | Ford Motor | 457 |
| 26V338 | Tiffin Motorhomes | 392 |
| 26V132 | ESCAPE TRAILER INDUSTRIES LTD. | 324 |
| 26V291 | Fontaine Trailer | 247 |
| 26V350 | Volvo Trucks North America | 205 |
| 26V010 | Gulf Stream Coach | 172 |
| 25V894 | Mercedes-Benz USA | 169 |
| 26V057 | Grand Design RV | 109 |
| 26V498 | Winnebago Industries | 96 |
| 26V289 | General Motors | 66 |
| 26V188 | Nissan North America | 51 |
| 26V001 | Volvo Car USA | 40 |
| 25V664 | REV Recreation Group | 39 |
| 25V342 | Battle Motors | 34 |
| 25V835 | Volkswagen Group of America | 27 |
| 26V428 | Crestline Coach Ltd. | 23 |
| 25V724 | Wartech Engineering | 18 |
| 25V571 | Orange EV | 16 |
| 26V264 | REV Recreation Group | 12 |
| 25V832 | Virnig Manufacturing | 10 |
| 25V482 | Hyundai Motor America | 10 |
| 26V431 | Kia America | 8 |
| 26V432 | Hyundai Motor America | 6 |
| 25V716 | Ferrari North America | 6 |
| 25V500 | The Heil Co. | 5 |
| 25V308 | Ford Motor | 3 |
| 25V331 | Hyundai Translead | 2 |
| 25V260 | Ford Motor | 2 |
| 25V290 | Prevost Car (US) | 1 |
Finding 2: these are physical failures, not software ones
Only 5 of the 60 campaigns describe a remedy involving software, firmware, reprogramming or recalibration. The other 55 require parts and a workshop.
This is the mirror image of what we found when we looked at over-the-air recall remedies. Software is arriving in the recall system, but it is arriving in the routine recalls. The campaigns urgent enough to tell you to stop driving are overwhelmingly the ones an update cannot touch.
Finding 3: it is not mainly an electric-car story
Battery and propulsion-system defects account for 14 of the 60 campaigns: a substantial share, and the reason EV fire risk gets the coverage it does. But they account for 352,595 vehicles, about 15% of the total.
By NHTSA’s own top-level component categories:
| Component category | Campaigns |
|---|---|
| Electrical system | 17 |
| Service brakes | 6 |
| Equipment | 5 |
| Wheels | 4 |
| Suspension | 4 |
| Engine and engine cooling | 4 |
| Fuel system | 3 |
And by vehicles affected, the ranking changes completely:
| Component category | Vehicles |
|---|---|
| Steering | 1,078,250 |
| Electrical system | 582,350 |
| Seats | 462,869 |
| Service brakes | 102,113 |
| Fuel system | 70,285 |
The single largest “park outside” instruction in the set is not a battery at all. It is a steering pump.
What the two flags actually mean
Do not drive means the agency and the manufacturer judged the risk of continuing to drive unacceptable. Park outside almost always means fire risk: the vehicle can ignite while parked and switched off, so it should be kept away from structures until repaired. A campaign carrying park outside is telling you the danger exists when the car is doing nothing at all.
The limits of this study
- The sampling frame is ours, not the industry’s, and it is narrower than some public counts. This analyses the 60 most recent of the 62 flagged campaigns in our own holding, which began recording on 12 August 2026 and is built from campaigns we have read. Other trackers list substantially more active critical recalls, because they draw on a longer and differently-constructed frame. Our totals are therefore a floor, and the distribution, not the total, is the finding. We are not claiming to enumerate every such recall in the United States.
- United States only. Canada and the United Kingdom do not publish comparable structured urgency flags.
- “Potentially affected” is the authority’s own phrase and is frequently revised after filing, usually downwards. We keep the earlier figures, which is why the change register exists.
- This is a snapshot of open campaigns, not a rate. It says nothing about how quickly repairs are completed, which no authority in our research publishes.
- Software classification is by remedy wording, so a software remedy that avoids the usual vocabulary would be missed.
Method and sources
Campaign records are read directly from NHTSA and stored with the wording they carried at the time of reading. Flags, components and unit counts were read on 14 August 2026. Component categories are NHTSA’s own top-level classifications, not ours. The full live list is published at recalls that say do not drive, or park outside.
Questions this study answers
How many cars are under a do-not-drive or park-outside recall?
In the 60 United States campaigns analysed on 14 August 2026, 2,422,750 vehicles were potentially affected. Twenty-seven campaigns carried a do-not-drive instruction, 34 carried park outside, and one carried both.
Are most urgent recalls large?
No, and this is the most misleading thing about how they are reported. The median campaign covers 629 vehicles and 36 of the 60 cover fewer than 1,000. Four campaigns account for 85% of all affected vehicles, and a single Chrysler steering-pump recall accounts for 44%.
Why does a recall tell me to park outside?
It almost always means fire risk. The vehicle can ignite while parked and switched off, so it should be kept away from buildings until the repair is done. The danger exists when the car is doing nothing at all.
Are do-not-drive recalls mostly about electric cars?
No. Battery and propulsion defects account for 14 of the 60 campaigns and about 15% of affected vehicles. By vehicle count the largest categories are steering, electrical systems and seats. The single biggest park-outside campaign in the set is a hydraulic steering pump.
Can an urgent recall be fixed with a software update?
Rarely. Only 5 of the 60 campaigns describe a software, firmware, reprogramming or recalibration remedy. The other 55 need parts and a workshop visit.
How to cite this study
CarsMultiverse (2026). The recalls that tell you to stop driving are mostly tiny. Published 14 August 2026. https://carsmultiverse.com/research/do-not-drive-recalls-analysed/
The figures were read on the date stated and are held with the wording they carried at the time. If they are restated, the change is noted on the page rather than made silently, so a citation to this version remains checkable.
Underlying records: the changes as JSON · RSS feed · the register itself.
Use this
Free to quote and republish with credit to CarsMultiverse and a link to this page. Every campaign number is a live NHTSA campaign and can be checked directly.
Published 14 August 2026. Figures were read on that date and will be restated, not silently edited, when this study is updated.