Key facts
- Finding: Larger recalls are repaired less often. Under 100,000 units: 72.8% mean / 81.9% median completion. Over 1,000,000 units: 63.5% mean / 67.7% median.
- Robustness: On the 2020-2024 cohort (old enough to be fixed), the small-vs-mega mean gap grows from ~9.3 to ~10.7 points, so the pattern survives the "too new to be fixed" objection.
- Dataset: N = 10,494 NHTSA vehicle recall campaigns; report dates 2011-05-25 to 2026-06-15; retrieved 2026-08-12.
- News peg 1 — Ram 1500 (26V495000): 1,271,294 trucks, MY2019-2026; second-row driver-side and center seat-belt buckle anchors may not be secured to the body; dealer reattaches, free; owner letters began ~Aug 18, 2026; no injuries reported.
- News peg 2 — Toyota Camry Hybrid (26V511000): ~508,354 vehicles, MY2025-2026 with the 7.0-inch cluster; software fault can leave the display blank at startup (no speedometer, fuel gauge, warnings) and disable turn/hazard signals; dealer software update, free, not over-the-air.
- Neither news-peg campaign is in the frozen file (it ends 2026-06-15), so both have effectively zero measured completion — they are entering the used pipeline before any repairs have accrued.
Do dealers have to fix recalls before selling a used car?
No federal law requires it. Franchised new-car dealers are barred from selling a new vehicle with an open recall, but there is no equivalent federal rule for used cars — independent lots and private sellers can legally sell a used car with an unrepaired safety recall. That means the responsibility to confirm the repair falls on the buyer, not the seller. Because the free remedy transfers with the car, you can still get it fixed after purchase, but you should never assume it was already done.
This is exactly where recall completion data matters. A recall being issued is not the same as a recall being repaired, and the gap between the two is large and predictable by size.
Does recall size predict whether it gets fixed? Yes — bigger means less-fixed.
Recall size predicts completion, and the relationship runs the opposite way to intuition: the bigger the recall, the smaller the share of vehicles actually repaired. Across all 10,494 vehicle campaigns in our file, median completion falls monotonically from 81.9% for sub-100,000 recalls to 67.7% for recalls over a million units. The all-years means decline as a clear gradient (one adjacent pair, 250k-500k vs 500k-1M, is effectively tied within a tenth of a point), and on the 2020-2024 cohort the means are strictly monotonic.
| Recall size (units affected) | All years: n | Mean % | Median % | 2020-2024: n | Mean % | Median % |
|---|---|---|---|---|---|---|
| Under 100,000 | 9,662 | 72.8 | 81.9 | 4,050 | 76.5 | 86.9 |
| 100,000–250,000 | 385 | 70.6 | 78.2 | 157 | 75.7 | 82.3 |
| 250,000–500,000 | 250 | 66.6 | 72.0 | 90 | 72.0 | 79.9 |
| 500,000–1,000,000 | 109 | 66.7 | 71.0 | 38 | 71.2 | 76.8 |
| 1,000,000+ | 88 | 63.5 | 67.7 | 25 | 65.8 | 70.3 |
The most common objection is that big recalls simply need more time. We tested that directly: restricting to campaigns reported in 2020-2024 — old enough that repairs should have landed — does not close the gap, it widens it. The mean spread between the smallest and largest bins grows from about 9.3 points (all years) to about 10.7 points (2020-2024 cohort). Whatever drives low completion on mega-recalls — scale, parts backlogs, resold cars, owners who moved — is structural, not a timing artifact. This tracks with what we found earlier looking at car recall completion rates across the very largest campaigns.
By the numbers
- 63.5% — mean completion for recalls over 1,000,000 vehicles (median 67.7%).
- 72.8% — mean completion for recalls under 100,000 vehicles (median 81.9%).
- ~10.7 points — the small-vs-mega mean gap on the 2020-2024 cohort, wider than the ~9.3-point all-years gap.
- 10,494 — vehicle recall campaigns tabulated.
Based on N = 10,494 NHTSA vehicle recall campaigns (recall_type=Vehicle, potentially_affected>0), from the frozen file nhtsa-recall-completion-2026-08-12.csv, retrieved 2026-08-12; completion_rate capped at 100% (only 3 tiny sub-100k campaigns exceed it, so the 1M+ aggregate is 63.51% capped or uncapped); "2020-2024 cohort" is binned on report_received_date year as an age proxy, not on vehicle model year.
Methodology
Anyone should be able to reproduce this from the same file, so here is exactly what we did.
Source. NHTSA's recall-completion flat file, retrieved 2026-08-12 and frozen unchanged as nhtsa-recall-completion-2026-08-12.csv. NHTSA replaces that dataset daily, so a live pull will not match ours; the frozen copy is the one every figure on this page was computed from.
Filters. Kept rows where recall_type = Vehicle and potentially_affected > 0, and where a completion rate was present. That leaves N = 10,494 campaigns.
Binning. Campaigns were grouped by potentially_affected into five size bands (<100k, 100k-250k, 250k-500k, 500k-1M, 1M+), and we report the unweighted mean and the median completion rate within each band. We report the median alongside the mean because completion is left-skewed and a handful of near-zero campaigns drag the mean.
Capping. · completion_rate was capped at 100%. Only 3 campaigns in the file exceed it, all small, and the 1M+ figure is 63.51% whether capped or not — so the cap does not carry the finding.
The robustness cut. The obvious objection is that large recalls are simply newer and have had less time to be repaired. We tested it directly by restricting to campaigns whose report_received_date falls in 2020-2024 (n = 4,360), old enough that repairs should have landed. The gradient does not close — it becomes strictly monotonic and the small-vs-mega mean gap widens from 9.2 to 10.8 points. We report this either way; had it reversed, this article would say so.
Known limits. · report_received_date is an age proxy, not vehicle model year. Completion rates are manufacturer-reported to NHTSA. And a completion percentage describes a campaign, not any individual car — which is exactly why a VIN-level check matters and a national average does not tell you about the car in front of you.
Which real mega-recalls are the least-completed?
The clearest examples are the airbag and inflator campaigns, and several sit far below even the 1M+ average. The single largest campaign in the file, Honda's 6,281,043-unit driver frontal airbag recall (15V320000), reached 79.0% — but that is the exception. GM's 5,877,718-unit ignition-switch recall (14V400000) sits at 37.1%, and Takata-related "Do Not Drive" and inflator campaigns cluster in the 35-47% band, with one Honda inflator recall (20V026000, 1,439,841 units) at just 7.9%.
| NHTSA campaign | Vehicles | Completion % | Maker / issue |
|---|---|---|---|
| 15V320000 | 6,281,043 | 79.0 | Honda — driver frontal airbag may rupture |
| 14V400000 | 5,877,718 | 37.1 | GM — ignition switch may turn off |
| 16V352000 | 4,648,270 | 43.5 | FCA — Do Not Drive, passenger airbag inflator |
| 15V313000 | 4,077,169 | 47.1 | FCA — Do Not Drive, driver airbag inflator |
| 21V158000 | 2,602,668 | 35.2 | Ford — driver airbag inflator may explode |
| 20V026000 | 1,439,841 | 7.9 | Honda — inflators may rupture/underinflate |
| 26V495000 | 1,271,294 | not yet in file | Ram 1500 — 2nd-row seat-belt buckle anchor (news peg) |
| 26V511000 | 508,354 | not yet in file | Toyota Camry Hybrid — 7.0-in cluster software (news peg) |
A Takata "Do Not Drive" recall is the most severe class NHTSA uses: the defect is dangerous enough that owners are told to stop driving until repaired, yet completion still stalls at 37-47% for these campaigns — evidence that even maximum-urgency messaging does not guarantee the fix reaches the car. If you are buying a vehicle from that era, the airbag is worth specific scrutiny; here is how to check if a used car had an airbag deployed and whether prior deployment or an open inflator recall is in play.
Why the fresh Ram 1500 and Camry Hybrid recalls are the ones to verify
Both new campaigns are large, recent, and — critically — not yet in any completion dataset, which is precisely why a buyer should treat them as unresolved. The Ram 1500 recall (26V495000) covers 1,271,294 trucks from model years 2019-2026: at the factory, the second-row driver-side and center seat-belt buckle anchors may not have been properly secured to the body structure, so a belt might not restrain an occupant in a crash. Stellantis opened its internal investigation on Feb 26, 2026 after a Transport Canada customer-concern report; no injuries have been reported. The dealer inspects and, if needed, reattaches the anchors free of charge, with owner letters starting around Aug 18, 2026.
The Camry Hybrid recall (26V511000) covers about 508,354 MY2025-2026 cars fitted with the 7.0-inch combination meter (higher trims with the 12.3-inch display are excluded). A software fault can leave that cluster blank at startup — no speedometer, fuel gauge, or warning lamps — and can also deactivate the exterior turn signals, hazard lamps, seat-belt warning, and smart-key reminder, which fails Federal Motor Vehicle Safety Standards. The remedy is a dealer software update: free, but not over-the-air, so the car must physically visit a dealer. (Campaign 26V511000 is confirmed directly against NHTSA's campaign record; owner notification letters are expected to be mailed September 21, 2026. Toyota's internal numbers are 26TB16 and 26TA16.)
Because both recalls are days old, huge numbers of these vehicles will change hands before repairs are logged. That is the entire lesson of the completion data in miniature: the freshest, biggest, most-publicized recall is the one least likely to already be handled on the specific car in front of you.
Is it safe to buy a used car with an open recall?
It can be safe once the free remedy is completed, but an open safety recall means a known defect is currently unrepaired on that vehicle. Treat any flagged recall as unresolved until you have proof otherwise: a repair-order line item, a dealer confirmation, or a clean result from NHTSA's VIN tool. The repair is free to you as the new owner, so an open recall is a reason to schedule the fix, not necessarily to walk away — but it is a hard reason to verify before you hand over money. If you want a worked example, see our guidance on whether you should buy a used Jeep Wrangler with an open recall and how to weigh a park-outside campaign.
What a VIN check can and can't tell you here
A VIN check tells you a car's recall exposure — that it falls within one or more campaigns — but not the remedy status. This distinction is the whole point of this story: because mega-recalls are the least-completed, a flagged recall must be treated as potentially open until confirmed. Use the free tool at nhtsa.gov/recalls first, and the fastest way to learn how to check a used car for open recalls before buying is to run the VIN there and then call a franchised dealer of that brand for the repair-order history.
| Question | VIN recall screen | Authoritative confirmation |
|---|---|---|
| Does this car fall within a recall? | Yes — presence/count | NHTSA VIN lookup / dealer |
| Was the recall actually repaired? | No | NHTSA VIN lookup / dealer repair order |
| Open-vs-closed / completed status? | No | NHTSA VIN lookup / dealer |
| Per-unit firmware / remedy detail? | No | Franchised dealer |
| Is there an open NHTSA investigation (PE/EA)? | No | NHTSA (nhtsa.gov) |
Where a broader history check earns its keep is on the records NHTSA's recall tool does not carry: accident and damage records with airbag-deployment status, odometer and rollback checks, junk and salvage auction records, theft (NICB) history, ownership history, and past and current sales-listing history with prices and mileage.
A Zilocar VIN check covers exactly that history layer and screens for recall presence and count — but it does not confirm a recall was remedied, does not track open-versus-closed status, and does not flag NHTSA investigations. For a mega-recall like the Ram 1500 or Camry Hybrid, use the VIN screen to know the car is exposed, then confirm the actual fix with the dealer or NHTSA.
