How Many Used Cars Are Under a Do-Not-Drive Recall? The 2026 Census

Key facts

• 108 active do-not-drive recall campaigns on cars and light trucks as of 2026-09-15, covering 20,752,140 vehicles (potentially-affected units at launch).
• 202 campaigns and 20,908,823 units across NHTSA's whole vehicle class — but 94 of those 202 campaigns are not cars or light trucks (motorcycles, buses, heavy trucks, motorhomes, trailers, industrial equipment). They are 46.5% of the campaign count and only 0.75% of the units, which is why the unit total barely moves and the campaign count nearly halves.
• 99.4% of car and light-truck do-not-drive units — 20,622,908 across 65 campaigns — are on vehicles whose newest affected model year is 2016 or older (10+ model years old). On the all-vehicle basis it is 98.7%.
• Airbag-inflator campaigns, overwhelmingly Takata, account for 50 campaigns and 95.5% of car and light-truck units (19,820,917). Strip every airbag campaign out and 58 car/light-truck campaigns covering 931,223 units remain.
• Even with every airbag campaign removed, 86.1% of the remaining car and light-truck units are on 10+ year-old vehicles — the old-car concentration is not just an airbag story.
• Largest single campaign: 16V352000, Chrysler/Dodge/Jeep/Ram/Mitsubishi passenger Takata inflator, MY2004-2012, 4,648,270 units.
• Largest non-airbag campaign: 21V473000, Cadillac/Saab rear-suspension arm may fracture, MY2010-2016, 380,498 units.
• A do-not-drive advisory means stop driving until repaired; the repair is free at a franchised dealer.

How do I check if a used car has a do-not-drive recall before I buy it?

Enter the VIN at nhtsa.gov/recalls (free) or run a VIN history report like Zilocar's — both surface whether an open recall exists on that specific vehicle. Neither lookup labels a recall as "do-not-drive" in a single click, and neither shows whether the repair was already completed on that VIN. To confirm a do-not-drive advisory and its open-versus-remedied status, read the recall's consequence and remedy language on NHTSA's page for the campaign, then call a franchised dealer of that make with the VIN.

The screening step matters most on old, cheap cars, because that is exactly where do-not-drive risk is concentrated. The same VIN-screening logic applies to the broader airbag class — see our guide on how to check if a used car has an open airbag recall for the Takata-heavy Dodge, Chrysler, and Jeep lineup that dominates this census.

How many used cars are under a do-not-drive order right now?

As of the 2026-09-15 data pull, 108 active do-not-drive campaigns on cars and light trucks cover 20,752,140 vehicles. "Active" here means the campaign carries a do-not-drive flag in NHTSA's current flat file; recall campaigns do not expire, so every one of them remains in force for any VIN that was never repaired. Two honest caveats: this is potentially-affected units at launch, not cars still on the road or still unrepaired today, and the count is of campaigns flagged do-not-drive, not of individual VINs.

One thing to be precise about, because it changes one of the two headline numbers and not the other. NHTSA's RCLTYPECD='V' class means "vehicle", not "car": it also carries motorcycles, buses, heavy trucks, motorhomes, trailers and even towable industrial equipment. On the whole class there are 202 do-not-drive campaigns covering 20,908,823 units. We classified all 202 by make and model, and 94 of them — 46.5% — are not cars or light trucks: 13 Yamaha and 9 Kawasaki motorcycle campaigns, 9 Mack and 4 Volvo heavy-truck campaigns, transit and school buses from Nova Bus, IC Bus, Gillig and New Flyer, Newmar and Thor motorhomes, Doosan air compressors, and more.

Those 94 campaigns hold just 156,683 units, 0.75% of the total. So the unit figure is robust — 99.25% of the 20.9M really is cars and light trucks — while the campaign count nearly halves, from 202 to 108, once you restrict it to what a used-car buyer could actually be shopping for. This page leads with the car and light-truck figures throughout, and the frozen CSV carries a vehicle_class column so anyone can reproduce either basis. (Watch one trap in that column: the make ESCAPE is Escape Trailer Industries, a travel-trailer builder, not the Ford Escape.)

The 2026 do-not-drive census — headline numbers

Based on 13,616 distinct NHTSA vehicle recall campaigns, filtered to the 202 flagged do-not-drive and then classified by vehicle type, from the ODI recalls flat file retrieved 2026-09-15 (max report date 2026-09-10). The right-hand column is NHTSA's unfiltered "V" class and includes motorcycles, buses, heavy trucks, motorhomes, trailers and industrial equipment; the bolded left column is the used-car answer.

Why would a 10- or 15-year-old car still be under a do-not-drive warning?

Because recall campaigns never close, and the volume of do-not-drive risk lives almost entirely in old cars. Sort the 108 car and light-truck campaigns by the newest model year they affect, and the pattern inverts: 43 of them are on 0-to-9-year-old vehicles, but those hold just 0.6% of the units. The other 99.4% of units are on cars 10+ model years old — mostly 2004-2016 Takata airbag vehicles that were sold, resold, and in many cases never brought in for the free inflator swap. (On NHTSA's whole vehicle class the same split is 1.3% against 98.7%.)

This is why the affordability end of the used market carries outsized do-not-drive exposure: the cheapest listings are disproportionately the model years where these orders concentrate. It parallels what we found enumerating how to check if a used car has a park outside fire recall — another structured NHTSA advisory whose volume skews toward older vehicles.

Is this just Takata? (The robustness split)

No — but airbag campaigns dominate the raw unit count, so we report both. One precision point first: our screen selects campaigns whose NHTSA component label contains AIR BAG, which is slightly broader than "Takata". It is the right screen anyway, because several manufacturers' Takata campaigns (BMW's and Mazda's among them) carry only the generic AIR BAGS label with no inflator wording, so an inflator-only text screen would drop them. Read the figures below as airbag-component campaigns, overwhelmingly Takata rather than as a certified Takata count.

Airbag-component campaigns: 50 of the 108 car and light-truck campaigns, holding 19,820,917 units (95.5%). Strip them out and 58 campaigns covering 931,223 units remain. Crucially, the "do-not-drive is an old-car problem" finding survives the strip: of the non-airbag campaigns, 86.1% of units are still on 10+ year-old vehicles. The two largest non-airbag do-not-drive campaigns make the point — a 2010-2016 Cadillac/Saab rear-suspension-arm fracture (380,498 units) and a 2006-2012 Mercedes-Benz brake-booster power-assist loss (350,294 units), both squarely in the used-car age band.

Top makes by do-not-drive units

Grouped by the manufacturer that filed the recall (NHTSA's MFGTXT), restricted to cars and light trucks. We group on the filer rather than on the make label on purpose: a single campaign often spans several makes — 16V352000 alone covers Chrysler, Dodge, Jeep, Ram and Mitsubishi — so assigning each campaign to one make splits manufacturers across rows and undercounts their campaign totals. The same file also spells one manufacturer two ways, MERCEDES BENZ and MERCEDES-BENZ, which a make-level table reports as two companies.

Largest individual do-not-drive campaigns

Tested against the full flat file, it is defensible to say airbag campaigns account for the majority (95.5% of car and light-truck units) of do-not-drive exposure, and that Chrysler (FCA US) files the most units of any single manufacturer. We do not claim any "first," and we do not rank models: POTAFF is a launch-time manufacturer estimate, not a count of cars still unrepaired, so a model-level "worst" list is not supportable from this file.

By the numbers

• 108 active do-not-drive campaigns on cars and light trucks; 20,752,140 vehicles covered.
• 202 campaigns and 20,908,823 units on NHTSA's whole vehicle class, of which 94 campaigns (0.75% of units) are motorcycles, buses, heavy trucks, motorhomes, trailers and industrial equipment.
• 99.4% of car and light-truck units are on 10+ year-old vehicles (MY ≤2016); still 86.1% after removing every airbag campaign.
• 50 airbag-component campaigns, overwhelmingly Takata = 95.5% of car and light-truck do-not-drive units.
• 13,616 distinct vehicle campaigns processed to isolate the 202.
• 0.6% of car and light-truck units are on vehicles 0-9 years old, despite those being 43 of the 108 campaigns.

Method in brief: NHTSA ODI recalls flat file FLAT_RCL_POST_2010 (tab-delimited, 29 fields), retrieved 2026-09-15, max report_received_date 2026-09-10. Filter RCLTYPECD='V' AND the structured do-not-drive advisory field = 'Yes'; deduplicate to distinct campaign number; POTAFF taken once per campaign; then classify by vehicle type. Full method below. Frozen filtered output, all 202 campaigns with a vehicle_class column: do-not-drive-stop-drive-recall-census-used-car-vin-check-2026-2026-09-15.csv.

A methodology note reporters can reuse

"Do not drive" is no longer an unstructured text field to guess at. NHTSA added structured DO_NOT_DRIVE (field 28) and PARK_OUTSIDE (field 29) columns to the recalls flat file in May 2025, and that flag is authoritative and reimplementable. We cross-checked it against naive text-matching for phrases like "do not drive," "stop driving," and "should not drive." Exactly how many that finds depends on which free-text fields you search: scanning the defect, consequence and remedy fields returns 12 campaigns and overlaps the structured set on 5, missing 197 of the 202; a wider scan returns 19 and overlaps on 9. Either way the conclusion is the same and it is not marginal — text-matching misses roughly 195 of the 202 campaigns. Takata's own remedy text reads "advised not to drive their vehicle," which evades those patterns, and the formal order often lives in a separate NHTSA consumer advisory. Use the structured flag; text-matching badly undercounts.

Method

Source. NHTSA's Office of Defects Investigation recalls flat file, FLAT_RCL_POST_2010.txt inside FLAT_RCL_POST_2010.zip. Retrieved 2026-09-15; the archive member's own timestamp is 2026-09-15 07:02 UTC; 311,124,477 bytes uncompressed; tab-delimited, 29 fields, no header row; 245,599 rows. CRC verified (testzip() returned clean), so the file is not a truncated download.

Freshness, asserted before any arithmetic. The newest RCDATE (recall report date) inside the file is 2026-09-10, five days before the run. We check this every time and abort rather than annotate if it fails, because NHTSA's complaints download silently regressed twice in six days — once serving records that stopped in 2018, once a 1995-1996 slice — each time under a normal-looking Last-Modified. A stale-but-parseable file is the failure mode, not a missing one.

Universe. · RCLTYPECD = 'V' (vehicle recalls), deduplicated to distinct CAMPNO: 13,616 campaigns.

Filter. Of those, the ones whose structured do-not-drive advisory field is Yes: 202 campaigns. That field is the 28th of the 29 tab-separated fields; the 29th is the park-outside advisory. We did not assume which was which from position. We split the campaigns by each flag and checked them against NHTSA's own API: 12V336000 and 13V103000 return parkIt=true, parkOutSide=false; 15V335000 and 18V601000 return the reverse; 22V541000 returns both true. That pins the do-not-drive column unambiguously, and it is the single check this whole census rests on — read the wrong column and the page is about park-outside advisories instead.

Aggregation. · POTAFF (potentially affected units) is taken once per campaign, never summed across a campaign's component rows — summing would multiply-count any campaign filed against several components. All 202 campaigns' rows agreed on POTAFF, so this choice is unambiguous rather than a judgement call.

Vehicle-type classification. NHTSA's V class means "vehicle", not "car": it includes motorcycles, buses, heavy trucks, motorhomes, trailers and towable industrial equipment. We classified all 202 campaigns from MAKETXT/MODELTXT into cars-and-light-trucks (108 campaigns, 20,752,140 units) and everything else (94 campaigns, 156,683 units). This classification is ours, not NHTSA's, and the frozen CSV exposes it as a vehicle_class column so anyone can redo or reject it. Two label traps we hit: make ESCAPE is Escape Trailer Industries, not the Ford Escape, and MERCEDES BENZ and MERCEDES-BENZ are one manufacturer under two spellings.

Manufacturer grouping. Where we report by company we group on MFGTXT, the Part 573 filer, not on MAKETXT. A single campaign frequently spans several makes (16V352000 covers Chrysler, Dodge, Jeep, Ram and Mitsubishi), so a make-level table splits one manufacturer across rows and undercounts its campaign total.

Reversal checks run, and what they changed.

1. The vehicle-type split is the check that moved a headline. Restricting to cars and light trucks barely touches the unit total (20,908,823 → 20,752,140, a 0.75% fall) but nearly halves the campaign count (202 → 108). The unit figure is robust; the campaign count is not, and an earlier draft of this page reported 202 as a used-car number. Both bases are published side by side above.
2. Airbag strip. Removing all 50 airbag-component campaigns leaves 58 campaigns and 931,223 units, and the old-car concentration survives at 86.1%. The finding does not depend on Takata.
3. Structured flag versus text-matching. Searching the free-text defect, consequence and remedy fields for "do not drive"/"stop driving"/"should not drive" finds 12 campaigns and overlaps the structured set on 5, missing 197 of 202. The structured flag is the only defensible source; text-matching is not a viable substitute.

Known limits, all of which bound what this page can claim.

• POTAFF is a manufacturer's launch-time estimate of units potentially affected. It is not a count of cars still on the road, still unrepaired, or still in the US. Many have been repaired, scrapped or exported. Nothing here is a "cars currently dangerous" figure.
• Because of that, this file cannot support a model-level "worst car" ranking, and we do not publish one.
• Campaign counts are of campaigns, not VINs. One car can sit under several campaigns.
• "Active" means the do-not-drive flag is present in the current file; recalls do not expire, so campaign age says nothing about whether a given VIN was fixed.
• The vehicle-type classification is judgement applied to free-text make/model strings; borderline cases (light commercial vans, class 4-5 cabovers) are decided in the CSV and can be re-decided from it.

Reproducibility. The frozen output for this story — all 202 campaigns with units, manufacturer, makes, models, model-year span, components, airbag and park-outside flags and our vehicle_class — is published at do-not-drive-stop-drive-recall-census-used-car-vin-check-2026-2026-09-15.csv. NHTSA republishes the source file continuously, so a later pull will not reproduce these counts exactly; the frozen CSV is the reproducible artefact.

What a VIN check can and can't tell you here

A VIN check does real work at the screening step, but be honest about its limits. It confirms a recall exists on a car; it does not confirm the recall was fixed.

The differentiated value of a VIN history report is the record around the recall — because when the recalled part is the airbag, whether that airbag already deployed in a prior crash is exactly what you want to know. This mirrors the deployment-history checks we walk through for how to check if a used car had an airbag deployed.

Screen with a VIN check, then verify with NHTSA and the dealer. A Zilocar VIN check flags open-recall presence and surfaces accident, airbag-deployment, salvage-auction, odometer, theft, ownership, and sales-listing history — the record that tells you what a cheap older car has actually been through. It does not certify that a do-not-drive recall was remedied, track NHTSA investigations, or show firmware or title-brand status; confirm open-versus-fixed recall status through NHTSA's VIN tool and a franchised dealer.

About the September 2026 recall wave

For context only, and never part of the numbers above: the week's headlines included a Stellantis do-not-drive warning on roughly 225,000 older Chrysler, Dodge, Ram, and Jeep vehicles with unrepaired Takata inflators. Multiple outlets corroborate the ~225,000 figure and the Takata mechanism but date the warning to February 2026, not September; a September re-issuance is plausible but we could not verify it, so treat the date as unconfirmed. Two other same-week recalls are not do-not-drive orders: Ford's ~223,000 F-150 (MY2023-2027) fuel-tank-strap recall and Jeep's 328,381 Grand Cherokee (MY2021-2023) rear-coil-spring recall. Both are standard recalls with free repairs.

Related checks

• Used Toyota/Lexus hybrid do-not-drive recall: what a VIN check shows
• Ford Bronco Sport do-not-drive recall check by VIN
• Used Chevy Silverado, Tahoe & Escalade transfer-case do-not-drive recall

Sources

• NHTSA ODI Recalls flat file (FLAT_RCL_POST_2010) — the primary dataset this census is computed over; contains the structured DO_NOT_DRIVE and PARK_OUTSIDE fields.
• NHTSA RCL record layout / data dictionary — documents DO_NOT_DRIVE (field 28) and PARK_OUTSIDE (field 29), added May 2025.
• NHTSA free VIN recall lookup — screen a specific VIN for recall presence (not remedy status).
• data.gov catalog: ODI Recalls Flat File — dataset provenance and update cadence.
• Frozen filtered dataset for this story (202 campaigns, recount-ready).
• CBS News — Stellantis do-not-drive warning for ~225,000 Takata-equipped vehicles — news-peg context.
• Cars.com — Ford recalls 223,000+ F-150 pickups over fuel-tank risk — same-week context (not a do-not-drive order).
• WSOC-TV — Jeep recalls 328,381 vehicles over rear coil springs — same-week context (not a do-not-drive order).

Last verified: 2026-09-15