Automotive Supplier Recalls Tracker: What Sets the Recall Population
In short: An automotive supplier recall is as large as the production record lets it be. Under US rules the manufacturer must tell NHTSA how many units may contain the defect and on what basis it chose them, so a supplier that can bound the bad window by station, shift, date and lot can keep the recall to that window, and one that cannot has to take in everything built since the last point it can prove was good.
- NHTSA received 115 recalls from 1 October to 12 November 2026. By the defect each summary states, 38 describe an assembly or process escape (a part missing, loose, not tightened or wrongly fitted), with a median of 277 units per recall, against 2,688.5 for the 52 design or component failures (our sort).
- One missing snap ring in Meritor drivelines produced three Part 573 reports between 28 October and 11 November 2026: Cummins for 206 parts, Daimler Trucks North America for 3,894 trucks and PACCAR for 982. The supplier's window was one operator's 39 days at one assembly station.
- Example plant: a shift-level record keeps 3,960 trucks in scope instead of 24,768, which saves US$303,381 to US$453,073 of technician time at BLS wage rates.
Tracker, updated 16 November 2026. Next update: December 2026. Sources are listed at the end.
A recall starts as an engineering question and ends as a counting question: which units, built when, shipped where. For a Tier 1 or Tier 2 supplier the answer sits in the plant's own records, the same records that back OEE for automotive Tier 1 suppliers: what ran on each station, on which shift and order, and when it stopped. This tracker reads the Part 573 safety recall reports that manufacturers file with the US National Highway Traffic Safety Administration (NHTSA), picks the cases where a supplier's part or process is named, adds a contrast case where no production record could narrow the scope, and shows how each report set its population. New cases go to the top of the table each month.
Supplier recall cases, October to mid-November 2026
Each row was read on 16 November 2026 in the Part 573 report itself. The first column is the date the report was received, as NHTSA's recalls dataset shows it; the production window is the one printed in the report, and for vehicles it is the vehicle build date.
| Received (dataset) | Recall | Filer and part | Units | Production window in the report | How the report says the population was set |
|---|---|---|---|---|---|
| 11 Nov 2026 | 26V743 | PACCAR: Kenworth trucks with Meritor drivelines that may be missing a snap ring | 982 | Vehicles built 4 Aug to 10 Sep 2026 | The dates product was received at its Ste. Therese and Chillicothe plants, plus the supplier operator's time at the station; clean point 11 Sep 2026 |
| 11 Nov 2026 | 26E078 | Hendrickson: STEERTEK NXT front steer axles whose Huck collars may not be fully swaged | 14,238 | Axles built 16 Jan to 10 Dec 2025 | Engineering, manufacturing and technical service information; the window ends at the corrective action of 10 Dec 2025 |
| 6 Nov 2026 | 26V737 | Daimler Trucks North America (DTNA): Freightliner and Western Star trucks with the same drivelines | 3,894 | Vehicles built 8 Jul to 22 Oct 2026, by model | Trucks fitted with drivelines identified by Cummins in a communication received on 26 Oct 2026 |
| 29 Oct 2026 | 26V718 | General Motors: Chevrolet Express and GMC Savana, a steering gear nut not properly tightened (gear from Nexteer) | 12,728 | Vehicles built 21 Jan 2025 to 1 Jun 2026 | Manufacturing records, part trace data and service records; a second look for 108 serial numbers left out of recall 26V399 |
| 28 Oct 2026 | 26E077 | Cummins: Meritor drivelines with a snap ring that may be missing from the slip yoke assembly | 206 | Built 14 Jul to 21 Aug 2026 | Three known failures traced to one operator at one assembly station |
| 2 Oct 2026 | 26V641 | Blue Bird: electric school buses whose high voltage battery mounting bolts may be loose or missing | 1,590 | Buses built 11 Jul 2018 to 17 Jun 2025 | Every bus fitted with the XALT battery; the cause is a torque specification set too low |
The Cummins figure counts only the drivelines it sold as replacement parts; the ones it sold as original equipment are counted in the vehicle makers' reports. The case notes below take each report in turn.
What NHTSA's recall data shows, 1 October to 12 November 2026
The US Department of Transportation's open data portal carries NHTSA's recalls as the dataset Recalls Data, attributed to NHTSA; the portal's metadata shows the data last updated on 16 November 2026. As received by NHTSA, it holds 115 recalls with a report received date from 1 October to 12 November 2026: 104 for vehicles, 9 for equipment and 2 for tires, 98 of them in October.
We sorted the 115 by the defect each summary states, with one rule applied in this order, first match wins:
- Label or certification: the summary names a label, a placard, a certification label or a certification symbol.
- Software: it names software, programming, an error, or a module that stops responding or sends a false command.
- Assembly or process escape: it says a part may be missing, loose or loosening, not tightened, fastened or secured, installed incorrectly or without something, the wrong part or material, welds not made as specified, built incorrectly, or a manufacturing defect.
- Design or component failure: everything else, such as a part that may fail, crack, rupture, detach, overheat or be damaged.
| Stated defect type | Recalls | Vehicle / equipment / tire | Units potentially affected | Median units per recall |
|---|---|---|---|---|
| Design or component failure | 52 | 48 / 3 / 1 | 1,560,968 | 2,688.5 |
| Assembly or process escape | 38 | 34 / 4 / 0 | 82,621 | 277 |
| Software | 17 | 16 / 1 / 0 | 861,214 | 286 |
| Label or certification | 8 | 6 / 1 / 1 | 12,926 | 233.5 |
| All | 115 | 104 / 9 / 2 | 2,517,729 | 548 |
Source: our count of the dataset's defect summaries and its "potentially affected" field. The sort follows the summary's words, not a root cause, and 19 of the 115 summaries also state a failure to comply with a Federal Motor Vehicle Safety Standard or with the certification rule in 49 CFR Part 567.
Assembly escapes make up a third of the reports, 38 of 115, but they are small: together they hold 82,621 of the period's 2,517,729 units, or 3.3% (our arithmetic). Our reading of why: a missing or loose part is a build event, so it has a first and a last unit that a production record can find, while a design fault sits in every unit built to the design and its population follows the design or the part number. The sort is only a first cut. Blue Bird's summary reads as an escape (bolts loose or missing), but its report puts the cause in the torque specification, and the population covers nearly seven years of builds.
One defect, several reports
A supplier's part can surface in several reports at once. 49 CFR 573.3(f) explains why: when original equipment goes into the vehicles of more than one manufacturer, the equipment maker reports on the equipment item and each vehicle maker reports on the vehicles in which it was installed. Three defect wordings repeat across makers in this period.
| Defect wording in the summaries | Reports | Filers | Units | What links them in the records |
|---|---|---|---|---|
| Snap ring may be missing in the slip yoke assembly of the driveline | 3: 1 equipment, 2 vehicle | Cummins, DTNA, PACCAR | 5,082 | DTNA's report cites Cummins' equipment recall 26E077; PACCAR names Meritor/Cummins as the Tier 1 supplier |
| Remote fuel filter hoses may separate from the fitting | 10 vehicle | 10 vehicle makers, from DTNA and PACCAR to Gillig and Nova Bus | 33,794 | Four summaries name L9 diesel engines, three of them as Cummins L9; the other six do not name the engine |
| Power cord fuse in power seating assemblies may not give adequate circuit protection | 5: 1 equipment, 4 vehicle | Lippert, Alliance RV, Forest River, Heartland, Ember | 14,510 | Lippert's equipment recall 26E070; the four RV summaries name certain power seating assemblies |
Cummins' driveline report names six buyers of the drivelines as original equipment: DTNA, Volvo Trucks, International Motors, Grove Worldwide, PACCAR and Nova Bus. By the dataset's 16 November update, two of them had filed a vehicle report for the snap ring, DTNA and PACCAR. Where affected drivelines went into another maker's vehicles, 573.3(f) puts a report on that maker too, so the next update checks for more.
What a Part 573 report must say about the population
The rule is 49 CFR 573.6, the defect and noncompliance information report, in 49 CFR Part 573 (Defect and Noncompliance Responsibility and Reports). A report is due "not more than 5 working days" after a defect is determined to be related to motor vehicle safety, or a noncompliance is determined to exist. The first report must carry at least the manufacturer's name, the population and the defect; the rest follows once it is confirmed. The population items are the ones a plant's records have to answer:
- The basis, paragraph (c)(2). The report gives "the manufacturer's basis for its determination of the recall population" and explains how the recalled units differ from similar ones left out. Passenger cars are identified by make, line and model year, other vehicles by body style or type, each with the inclusive months and years of manufacture. Equipment is identified by its generic name, part number, brand and model, with the months and years of manufacture "if available".
- Who else is involved, (c)(2)(iv) and (v). A component made by another manufacturer is named with its maker, and an equipment maker names every manufacturer that bought the item for new vehicles. That is why Cummins' report lists six buyers.
- The count and the share, (c)(3) and (c)(4). The total number of units potentially containing the defect, and the estimated percentage that actually contain it. In the snap ring case those shares are 1% for Cummins and DTNA and 0.3% for PACCAR. The difference between a defect and a defective unit is set out in defect vs defective.
- The chronology, (c)(6). For a defect, the principal events that led to the decision, with warranty claims, field reports and their dates.
The same section requires amendments to the population and the count within 5 working days of new information that updates or corrects them, and the reports show it happening. Cummins' report, received on 28 October, was amended on 30 October and again on 4 November. GM's report for 26V718 counted 12,715 vehicles on 29 October and 12,728 in the amendment of 6 November.
The statute sets the duty behind the paperwork. Under 49 U.S.C. 30118(c) a manufacturer of a vehicle or replacement equipment notifies the Secretary of Transportation, owners, purchasers and dealers when it learns of a defect and decides in good faith that it is related to motor vehicle safety. Under 49 U.S.C. 30120(a)(1) the manufacturer "shall remedy the defect or noncompliance without charge", and for a vehicle it chooses between repairing it, replacing it with an identical or reasonably equivalent vehicle, or refunding the price less a reasonable allowance for depreciation. Neither section sets how a vehicle maker and its supplier share the cost; the reports show who runs the remedy, and in the driveline case Cummins will administer DTNA's recall and send the owner letters.
For a supplier this turns the population into a claim it must be able to defend in writing. The table maps each part of the report to the plant record that answers it; how lots, serials and orders join in that record is covered in the guide to lot traceability.
| Part of the report | What it asks | The plant record that answers it |
|---|---|---|
| Basis, (c)(2) | Why these units, and how they differ from similar units left out | Dated events at the station: stops with reasons, changeovers, tool or fixture changes, the fix; the shift roster for who worked there |
| Inclusive dates, (c)(2) | Months and years of manufacture | Good counts by station, shift and product, with times |
| Buyers, (c)(2)(v) | Each manufacturer that bought the equipment | Shipments by lot, date and ship-to plant |
| Count, (c)(3) | Units potentially containing the defect | The good count inside the window, less units proven good |
| Share, (c)(4) | Estimated percentage that actually contain it | Results of the containment sort and of field returns |
| Chronology, (c)(6) | Principal events with dates | Field reports, warranty claims, internal reviews and decisions |
Case notes, November 2026
Each note uses facts from the report itself. Where we read more into a report than it says, the text marks it as our reading.
Meritor drivelines: Cummins 26E077, DTNA 26V737 and PACCAR 26V743
Cummins' report, in its second amendment of 4 November 2026, covers Meritor drivelines of six model series, RPL10 to RPL35 and 176MXL. A snap ring may be missing from the slip yoke assembly, which can let components migrate and may end in the driveline separating from the vehicle. The report says the population "was established by tracing all three known driveline failures to a single operator" who worked at the affected assembly station from 14 July to 21 August 2026. The cause it gives is that operator's deviation from the documented assembly process. A customer found the first missing ring on 8 August, more were found at a vehicle maker's site on 12 August, and the corrective actions of 21 August added operator training, enhanced in-process inspections and snap ring installation verification at three more production stations.
The two vehicle makers drew their windows from their own records. DTNA's population, 3,894 trucks of seven models, uses the drivelines Cummins identified, with build dates from 8 July to 22 October 2026; its report says none of the potentially affected drivelines remain at DTNA facilities. PACCAR's 982 Kenworth trucks were built from 4 August to 10 September 2026, a window based on the dates product was received at its plants in Ste. Therese and Chillicothe. PACCAR found its first missing ring at Ste. Therese on 8 August, ran two rounds of plant inspections and gives its clean point as 11 September 2026.
- The window is a roster window. The 39 days from 14 July to 21 August are one person's time at one station (our arithmetic for the count of days). Our reading: tracing failures to one operator's dates takes a record of who worked which station on which days, and that record set both edges. Without it the report would have needed another edge, such as the last audit that proved the station good.
- The customer's window is a receiving window. PACCAR's trucks start on 4 August and end at a clean point on 11 September, 21 days after the supplier's fix. Our reading: drivelines sat in stock between the supplier's line and the truck line, and the vehicle maker closed its window by inspecting its plants. A supplier's shipments by lot and date are what let a customer close that window earlier.
- One miss, three filings. Under 573.3(f), Cummins filed the equipment report, counting the 206 drivelines it sold as replacement parts, and each vehicle maker filed for its own trucks: 3,894 and 982. Two of the three reports state the cause; DTNA's cause field is blank and it quotes Cummins for the defect.
Hendrickson STEERTEK NXT axles, 26E078
Hendrickson's report covers 14,238 STEERTEK NXT front steer axles built from 16 January to 10 December 2025, with an estimated 0.1% containing the defect. Each axle has two axle seats press-fitted onto the axle beam and secured with Huck fasteners, four in all, and one or more Huck collars may not have been properly swaged on their pins. In normal operation the axle stays locked in the assembly station until the system records four good swaging operations. During and after rework, which happens when an incorrect swage is recorded, an axle could move on without that confirmation. On 10 December 2025 Hendrickson changed the process so that the axle stays locked during any rework, until a line supervisor or quality engineer checks the collars and enters a password.
The chronology starts on 24 September 2026, when a review of warranty information showed four field reports of missing Huck pins on axles built from January to September 2025. A review on 1 October found five more axles, built from 20 January to 3 December 2025, that quality teams at Hendrickson or the vehicle makers had already contained. Because the corrective action took effect on 10 December 2025 and no issues have been found on axles built since, the suspect population runs from 16 January to that date. A temporary 100% check of the Huck fasteners before shipment started on 7 October 2026.
- The normal path was gated; the rework path was not. The station's own record of four good swages protected the normal flow. The escape route was rework, so the window runs back across nearly a year of builds for nine known axles, four from field reports and five already contained. The report does not say whether rework was logged per axle; our reading is that a rework log by serial number with a time would have let a plant argue for the reworked axles rather than the whole period.
- The fix date closes the window. The population ends on 10 December 2025 because that is when the stronger control took effect. An end date is only as good as the record that proves when the fix was made.
- Serial lists travel to the customers. Hendrickson planned to notify Kenworth, Peterbilt, International and Blue Bird by 16 November 2026 with the list of affected axle serial numbers for each. Shipments recorded by serial and customer are what turn an equipment population into vehicles.
GM Express and Savana steering gears, 26V399 and 26V718
The first recall, 26V399, submitted on 23 June 2026, covered 26,541 Chevrolet Express and GMC Savana vans. A malfunctioning part sensor at the supplier (the component maker in the report is Nexteer Automotive Group) let some steering gear assemblies with only hand-tightened nuts pass inspection. The supplier reviewed its torque trace data to find the suspect gears; GM's later report, 26V718, says dealers checked vehicles for 51 suspect serial numbers. The condition was contained at the supplier on 29 May 2026, and vans built after 1 June 2026 have properly torqued nuts.
On 28 September 2026 a GM brand quality manager passed on a dealer's report of a loose nut on a 2025 Express that had already been inspected and closed under 26V399. GM matched its serial number to the supplier's list of suspect parts and found that 108 suspect serial numbers had been left out of the first recall because they "were believed to have been contained at the supplier". Recall 26V718 sends about 12,715 already-inspected vans back to dealers to check the gear serial against the longer list. Vehicles with a recorded non-suspect steering gear serial number, or a gear already replaced under 26V399, are excluded.
- Torque trace data found the gears; the containment record lost 108. The process record was good enough to name suspect gears by serial. The record that failed was the one saying where each suspect gear physically was.
- A recorded serial takes a van out of scope. 26V718 excludes vans whose gear serial was recorded as non-suspect. Where the serial was not recorded at the repair, the van goes back to the dealer.
- The same trigger twice. Both recalls began with a dealer's report of a loose nut that a GM brand quality manager passed on through GM's Speak Up for Safety program, on 28 May and on 28 September 2026. A closed recall is only as complete as the serial list behind it.
Blue Bird electric school buses, 26V641
Blue Bird's amended report of 5 October 2026 covers 1,590 Vision and All American electric school buses, model years 2018 to 2024 and 2026, all built from 11 July 2018 to 17 June 2025. The XALT high voltage battery may have loose, broken or missing mounting bolts. A field report of broken bolts in September 2023 led to a service manual change; in December 2025 a field report said a battery pack had dislodged from a bus; and in September 2026 Blue Bird concluded that the torque specifications set for the bolts at installation were insufficient. The report notes five field reports from September 2023 to May 2026, and says vehicles outside the population are not equipped with XALT HV batteries. A smaller Blue Bird recall in the dataset, 26V642 for 16 transit buses, uses the same summary wording.
- When the cause is the specification, time cannot narrow it. Every bus in the period fitted with the XALT battery is in scope, and the only difference the report gives from buses left out is that they do not have that battery. The population is bounded by a configuration record, which buses have the battery, not by a shift or a date.
- The summary and the cause differ. Loose or missing bolts read like an assembly escape, and our word sort counts them as one. The report shows the reason a sort by words is only a first cut: the bolts were fitted to a torque that was too low for the job.
- A service interval is not a production fix. The 2023 change asked for the bolts to be serviced at 12 months or 20,000 miles. The remedy now refits them at a new torque range with thread locker, and service moves to every 6 months or 10,000 miles. The battery is no longer offered in production.
The next update adds new cases under these, each with its report number and the date it was received.
Worked example: pricing the window at a retaining-ring station
Example plant. Every operating figure in this section is Example data; only the wage rates are sourced. The plant is not Meritor, Cummins or any company named above, and the sums are not an estimate of any recall in this tracker.
- The station: a Tier 2 cell presses a bearing into a yoke and fits an internal retaining ring, then ships the subassemblies to a Tier 1 that builds them into drivelines for heavy trucks. The ideal cycle is 60 seconds, so an 8-hour shift of 28,800 seconds makes at most 28,800 / 60 = 480 assemblies. The cell runs two shifts, five days a week.
- Normal output: 408 good assemblies a shift, so OEE is 408 x 60 / 28,800 = 0.85, or 85.0%.
- The event: at 19:40 on a Tuesday, second shift logs a stop at the machine with the reason "ring feeder jam". To keep running, the crew feeds rings by hand and switches off the ring-presence check, which sits on the feeder. It runs that way on its shifts; first shift runs with the feeder and the check on. A supervisor finds the check switched off on the eleventh second shift, and the restart with the check on is logged as a stop with its reason. While hand-fed, second shift averages 360 good assemblies: 360 x 60 / 28,800 = 0.75, or 75.0% OEE.
- The find: three weeks later, the Tier 1's receiving check finds a yoke with no ring. The last documented good audit of the station was 31 production days before the restart.
- The remedy: each assembly ends up in one truck. Every truck in the population gets an inspection of 0.6 technician hours, and trucks with a missing ring get a new driveline.
Step 1: the three populations. What the plant can prove sets the count.
- A, the stop log and counts by shift: second shift only, from the 19:40 stop to the logged restart. 11 x 360 = 3,960 assemblies, counting the first and the eleventh shift in full, which errs towards a larger population.
- C, dates only: the same eleven days, both shifts. 11 x 408 = 4,488 on first shift, plus 3,960, is 8,448.
- B, no event record: back to the last good audit, 31 days of two shifts, 62 shifts. 51 normal shifts x 408 = 20,808, plus 3,960, is 24,768.
Step 2: the wage. The remedy is done on heavy trucks, so the wage is the one for the people who work on them. In the BLS Occupational Employment and Wage Statistics for May 2025, bus and truck mechanics and diesel engine specialists (SOC 49-3031) earned US$24.30 an hour at the 25th percentile and US$36.29 at the 75th, nationally; the mean was US$30.92. These are wages before benefits and before any labour rate a dealer charges, so every sum below is a floor. One inspection costs 0.6 x 24.30 = US$14.58 to 0.6 x 36.29 = US$21.774, about US$21.77.
Step 3: hours and money per population.
| Population | Trucks | Inspection hours (x 0.6) | At US$24.30 an hour | At US$36.29 an hour |
|---|---|---|---|---|
| A: stop log and counts by shift | 3,960 | 2,376 | US$57,736.80 | US$86,225.04 |
| C: dates only | 8,448 | 5,068.8 | US$123,171.84 | US$183,946.75 |
| B: back to the last good audit | 24,768 | 14,860.8 | US$361,117.44 | US$539,298.43 |
The sums: 3,960 x 0.6 = 2,376 hours, 2,376 x 24.30 = 57,736.80 and 2,376 x 36.29 = 86,225.04; 8,448 x 0.6 = 5,068.8 hours, 5,068.8 x 24.30 = 123,171.84 and 5,068.8 x 36.29 = 183,946.752; 24,768 x 0.6 = 14,860.8 hours, 14,860.8 x 24.30 = 361,117.44 and 14,860.8 x 36.29 = 539,298.432.
Step 4: what the shift record is worth. Against the audit-only baseline, the shift record keeps 24,768 minus 3,960 = 20,808 trucks out of the inspections. 20,808 x 0.6 = 12,484.8 hours, so 12,484.8 x 24.30 = US$303,380.64 to 12,484.8 x 36.29 = US$453,073.39 of technician time for one event. Against a date-only record the gap is 4,488 trucks, 4,488 x 0.6 = 2,692.8 hours, or 2,692.8 x 24.30 = US$65,435.04 to 2,692.8 x 36.29 = US$97,721.71.
Step 5: what does not change. The trucks that really have no ring need a new driveline in every row. A narrower population saves only the inspections of good trucks, and only if the window truly holds every missing ring. GM's second recall, 26V718, shows the price of a list that leaves suspect parts out: about 12,715 vans went back for a second check.
Step 6: OEE flagged the window. Second shift's OEE fell from 85.0% to 75.0% for eleven shifts, 408 minus 360 = 48 assemblies a shift and 11 x 48 = 528 over the window, with the stop and its reason logged at the start. Our reading: a performance drop that starts at a logged stop and lasts several shifts is worth a quality question as well as a production one.
The money leaves out parts, freight, owner letters, the customer's own costs and the days its trucks are off the road, so it is a floor for one event on one station.
Book a demo and see losses by line and shift.
Bound, widen or prevent: choosing what to fund at the station
A shift record bounds the window after an escape; other controls stop the escape or shrink the window further. The table sets them side by side, with the anchors from the cases and the worked example.
| Option | Choose it if | Cost anchor on this page | Watch for |
|---|---|---|---|
| Rely on periodic audits | Volumes are low, or every part is already traced by serial number to its vehicle | Population back to the last good audit: 24,768 trucks, US$361,117 to US$539,298 of inspections (Example) | Every day between two audits is in scope |
| Record counts, stops and reasons by station, shift and order | Lots leave every shift, and an escape can start with a stop, a changeover or a crew change | 3,960 trucks, US$57,737 to US$86,225 (Example) | Each shift's output has to map to the lots it shipped |
| Gate every path past the check | A part can pass the check through rework, hand feeding or an override | Controls work, not priced here | Hendrickson's escape route was rework; its fix locks the axle until a supervisor or quality engineer releases it |
| Record part serials through to the vehicle | The part is safety-related and the customer records serials at build and in service | Labels and scans, not priced here | GM excluded vans with a recorded non-suspect serial; vans with no record were checked again |
| Keep a containment register | Suspect parts sit in your plant, in transit and at customers at the same time | Not priced here | GM's 108 serials were believed contained at the supplier and were not |
Two of these change the station itself. A check that cannot be bypassed is the idea behind poka-yoke, and the audit interval that sets the worst-case window is the subject of layered process audits. Whether a part needs a lot or a serial number is set out in lot vs serial tracking.
Questions to ask a vendor of line data or traceability software before you buy:
- Can you show good counts by station, shift and production order for any date range in the past year, and export them as a file?
- Is a stop stored with its start time, its duration and the reason the operator gave at the machine?
- Can a shipped lot be traced back to the shifts and orders that built it?
- Which paths past a check leave a record in your system: rework, hand feeding, overrides, a sensor out of service?
- Who can change a record after the shift, and is the change logged?
From field report to a population you can defend: the first week
The Part 573 clock starts at the determination, not at the first field report, so the plant's own work can start earlier. These steps fit a supplier plant, whether the plant files itself or feeds a customer's filing.
- Quality manager: write down the first known bad unit and the record that dates it: serial, lot, build date. Cummins' chronology starts with a customer's find on 8 August.
- Production manager: export good counts, stops and reasons by station and shift, from the last documented good audit to today, before anyone edits them.
- Quality and production together: find the edges. The earliest event that can explain the escape (a stop, a changeover, a new tool or fixture, a crew change on the roster) and the latest (the fix). Make each one a dated entry; it is the evidence the containment step of 8D problem solving relies on.
- Maintenance manager: list every path past the check: rework, hand feeding, overrides, a sensor out of service. Hendrickson's gap was the rework path.
- Materials: map the window's shifts to shipped lots, ship dates and plants. A customer may bound its own vehicles by the dates it received the parts, as PACCAR did.
- Quality manager: build the containment register: where each suspect unit is, with a name against each location.
- Plant manager: write the basis the way 573.6(c)(2) asks for it: why these units, how they differ from those left out, and the inclusive dates. Send each customer its share.
This quarter
- Run a trace on one press-fit or ring station: pick a random hour from last month and time how long it takes to name its lots and the plants they went to.
- Make every bypass of a check a logged stop with a reason, so the start of a window is a fact.
- Add the part serial to the shipping record for safety-related parts.
- Compare the interval between station audits with the population it implies, using the arithmetic in the worked example.
- Keep records findable for as long as the parts are in service, including after a line is moved or rebuilt. When a supplier closes a plant, the records that should travel with each tool are listed in the auto supplier plant closures tracker.
The shift record that bounds a window
Every case in this tracker turns on a record that existed or did not: a roster for one station, a receiving date at a truck plant, a rework path with no gate, a list of serials believed contained. The plant cannot choose its next escape, but it can choose which of those records it keeps. When the supplier itself is insolvent, the customer's choice to support, resource or buy ahead is worked through in the supplier insolvencies tracker. For the difference between following a lot and following a single part through its components, see traceability vs genealogy; for how OEE data can stand up in an audit, see audit-ready OEE traceability; and for the orders, routings and part records around the line, see MES for automotive suppliers.
Fabrico is a cloud OEE and light MES platform. It shows OEE and losses by machine, line and shift, with data from the machine's PLC or an IoT device with sensors on the machines you connect, and operators give the reason for a stop at the machine. That is a time-stamped record of what ran on each station, on which shift and order, and when it stopped, which is the record that bounds a window. With the selling price and margin of each SKU entered, losses are priced in money, so stops can be ranked by what they cost. Who worked the station comes from the shift roster or the plant's own records. On manual stations, AI cameras (an add-on) count output; they do not check parts, confirm that a ring is fitted or read labels. The building blocks are OEE software and a light MES for the production record.
Book a demo and see downtime by station and shift.
What people ask next
How is the population of a vehicle recall decided?
The manufacturer decides it and has to explain it. Under 49 CFR 573.6(c)(2), the Part 573 report gives the manufacturer's basis for the recall population, how the recalled units differ from similar ones left out, and the inclusive months and years of manufacture. In practice the basis is a record: Cummins traced its driveline failures to one operator's 39 days at one station, PACCAR used the dates the parts were received at its plants, and GM used part trace data and serial numbers.
Who files the recall when a supplier's part is defective?
It depends on who bought the part. Under 49 CFR 573.3(f), when original equipment is installed in the vehicles of more than one manufacturer, the equipment maker reports on the equipment item and each vehicle maker reports on the vehicles in which it was installed. That is why one missing snap ring produced three reports in this tracker: Cummins for 206 replacement drivelines, DTNA for 3,894 trucks and PACCAR for 982.
How fast must a Part 573 report reach NHTSA?
Not more than 5 working days after a defect is determined to be related to motor vehicle safety, or a noncompliance is determined to exist. The first report must give at least the manufacturer, the population and the defect, and the manufacturer must amend the population and the count within 5 working days after it has new information that updates or corrects them.
Who pays for the remedy in a vehicle recall?
Not the owner. Under 49 U.S.C. 30120 the manufacturer remedies the defect without charge, by repairing the vehicle, replacing it with an identical or reasonably equivalent one, or refunding the price less a reasonable allowance for depreciation. The statute does not set how a vehicle maker and its supplier share the cost. In the driveline case, DTNA's report says Cummins will administer the recall and send the owner letters.
Which production records narrow a recall population?
The ones that put dated edges on the window: good counts by station, shift and production order; stops with their times and the reasons operators gave; changeovers, tool changes and fixes as dated events; the shift roster; shipments by lot, date and ship-to plant; and part serials recorded at build and in service. In the Example plant on this page, a shift-level record cuts the population from 24,768 trucks to 3,960.
Where can I find automotive supplier recalls each month?
NHTSA's recalls are published as the Recalls Data dataset on the US Department of Transportation's open data portal, and each recall's Part 573 report is posted as a PDF on NHTSA's document server. Filter by component or manufacturer, then read the report for the supplier, the production dates and the basis for the population. This tracker does that each month for cases where a supplier's part is named.
Sources
Sources accessed on 16 November 2026 unless marked as accessed on another date. Dates are publication or submission dates where a page gives one; other dates are labelled as what they are.
- NHTSA, Part 573 Safety Recall Report 26V743 (PACCAR Incorporated), submitted 11 November 2026, accessed 16 November 2026
- NHTSA, Part 573 Safety Recall Report 26E078 (Hendrickson USA, L.L.C.), submitted 11 November 2026, accessed 16 November 2026
- NHTSA, Part 573 Safety Recall Report 26V737 (Daimler Trucks North America, LLC), submitted 6 November 2026, accessed 16 November 2026
- NHTSA, Part 573 Safety Recall Report 26V718 (General Motors, LLC), amendment 1, submitted 6 November 2026, accessed 16 November 2026
- NHTSA, Part 573 Safety Recall Report 26V718 (General Motors, LLC), first report, submitted 29 October 2026, accessed 16 November 2026
- NHTSA, Part 573 Safety Recall Report 26E077 (Cummins, Inc.), amendment 2, submitted 4 November 2026 (report received 28 October 2026), accessed 16 November 2026
- NHTSA, Part 573 Safety Recall Report 26V399 (General Motors, LLC), submitted 23 June 2026, accessed 16 November 2026
- NHTSA, Part 573 Safety Recall Report 26V641 (Blue Bird Body Company), amendment 1, submitted 5 October 2026, accessed 16 November 2026
- US Department of Transportation open data portal, Recalls Data (National Highway Traffic Safety Administration), data last updated 16 November 2026 (portal metadata), rows with report received dates 1 October to 12 November 2026, accessed 16 November 2026
- Cornell Law School LII, 49 CFR 573.6: Defect and noncompliance information report, last amended at 79 FR 43677, 28 July 2014, accessed 16 November 2026
- Cornell Law School LII, 49 CFR 573.3: Application, last amended at 68 FR 18142, 15 April 2003, accessed 16 November 2026
- US Government Publishing Office (govinfo), 49 U.S.C. 30118: Notification of defects and noncompliance, United States Code, 2023 Edition, accessed 16 November 2026
- US Government Publishing Office (govinfo), 49 U.S.C. 30120: Remedies for defects and noncompliance, United States Code, 2023 Edition, accessed 16 November 2026
- U.S. Bureau of Labor Statistics, OEWS series OEUN000000000000049303107: hourly 25th percentile wage, bus and truck mechanics and diesel engine specialists (49-3031), national, May 2025 estimates, accessed 16 November 2026
- U.S. Bureau of Labor Statistics, OEWS series OEUN000000000000049303109: hourly 75th percentile wage, bus and truck mechanics and diesel engine specialists (49-3031), national, May 2025 estimates, accessed 16 November 2026
- U.S. Bureau of Labor Statistics, OEWS series OEUN000000000000049303103: hourly mean wage, bus and truck mechanics and diesel engine specialists (49-3031), national, May 2025 estimates, accessed 16 November 2026
Latest from our blog

Real Fabrico tiles, sample data.



