On 7 of 28 Models, Engine Complaints Run Higher on the Model Year the Recall Doesn't Cover - in Every Exposure Window We Tested

Key facts

• 7 of 28 models show a higher engine-complaint share on the first model year the recall does not cover, at matched exposure, in every window tested: Ford F-250 Super Duty, Jeep Grand Cherokee, Jeep Renegade, Kia Telluride, Honda CR-V, Toyota Tundra, Chevrolet Silverado 1500.
• 19 of 28 models hold their sign across all five windows. 12 are consistently lower on the uncovered year. 9 are window-dependent.
• Toyota Tundra: 46 engine complaints of 134 on the uncovered MY2025 against 44 of 133 on the covered MY2024, in a matched 24-month window - the same rate at the same age, a gap of +1.2 to +4.2 points across windows.
• The obvious version of this analysis gives the opposite answer. On lifetime totals, 9 of 31 pairs flip sign; the Jeep Wrangler reads +25.9pp lifetime and −24.8pp matched.
• No worst-model ranking appears in this article, because magnitudes do not survive: the Kia Telluride spans +2.6 to +29.9 points on window length alone.
• A rise past a cutoff is a question, not a proven defect - the Honda CR-V's rise is a different fault from the one its recall addressed.

Why compare model years at all?

Because a recall is a line drawn around a population, and the car just outside it is often nearly the same car. The cutoff is the newest model year inside a filing's definition. Owners one year up get no letter and no free remedy, whether or not their car shares the part.

That makes one question answerable from public data: at the same age, do owners of the first uncovered model year report engine trouble at a higher rate than owners of the last covered one? NHTSA publishes both halves - the recall file carries every campaign's affected model years, the complaint file carries every filing with its date, model year and component - but nothing joins them, and no commercial vehicle-history aggregator publishes the join.

The seven models where the direction holds

Across 475 car and light-truck make/model combinations carrying at least one of 324 engine recall campaigns, 39 have an adjacent model-year pair measurable at matched exposure and 28 are measurable in at least three of the five windows. Seven of those 28 are higher on the uncovered year every time.

Counts are complaints filed in the first 24 months after 1 October of the preceding calendar year, so each pair is observed at the same age. The table is a list, not a ranking - see the window check below for why. The Ford F-250 Super Duty's uncovered year rests on only 35 complaints, so its percentage is the least stable of the seven even though its direction is not.

Is the Tundra the story the headlines say it is?

Directionally yes, in magnitude no, and the federal record is both firmer and narrower than the anecdote. The coverage this week rests on seven owners of 2025 Tundras reporting engine failures outside a recall that stops at model year 2024. The complaint file holds far more than seven: in a matched 24-month window the uncovered MY2025 Tundra logged 46 engine complaints out of 134 total, against 44 of 133 on the covered MY2024.

But that is parity, not excess - a gap of +1.2 to +4.2 points, among the smallest of the seven. The Tundra's claim to attention is that the direction is consistent, not that the gap is large.

One piece of context does belong to the Tundra, and it is why owners expect movement: the cutoff has already been redrawn twice. Campaign 24V381000 (May 2024, 102,092 units) covered 2022-2023 Tundra and Lexus LX600 with the V35A engine, for manufacturing debris that can contaminate the engine and fail a main bearing. 25V767000 (November 2025, 126,691 units) restated it and added model year 2024. 26V320000 (May 2026, 43,566 units) returned to MY2024 alone. A population widened once can be widened again - but whether it will be is Toyota's and NHTSA's decision, and nothing here predicts it.

Where the regulator is already looking past a cutoff

On the GM V8, visibly. The Chevrolet Silverado 1500's cutoff comes from 25V274000, covering 2021-2024 Escalade, Silverado 1500, Suburban, Tahoe, Sierra 1500, Yukon and Yukon XL fitted with the 6.2-litre L87 V8, for a bearing failure that can cause a loss of drive power.

NHTSA is looking beyond that cutoff itself. Its engineering analysis EA26005 covers roughly 997,743 L87-equipped vehicles spanning model years 2021 to 2026 - two model years past where the recall stops - and GM's deadline to produce engine-failure records, warranty data and its own internal assessment falls on 13 November 2026. So on this engine, the gap between the recall population and the complaint population is not a novel observation from a dataset; it is the subject of an open federal investigation. Whether an investigation becomes a recall is a separate question with its own base rates, and an open analysis is not a finding of defect.

Note the sibling models diverge: the Silverado 1500 is in the stable-higher seven while the GMC Sierra 1500, covered by the same campaign, is consistently lower on its uncovered year (−3.2 to −7.3 points). Same engine, same cutoff, opposite readings - which is itself a reason to distrust single-model storytelling from this file.

What the robustness checks killed

Two checks changed the answer, and both matter more than any individual row.

Check one: lifetime totals are confounded, and they reverse the result. The intuitive approach compares lifetime complaint totals. Do that and 9 of the 31 pairs measurable at 24 months flip sign.

Two mechanisms push the same way: the first uncovered model year is always the newer car, with less road time to accumulate filings, and the recall itself generates complaint traffic onto the years it covers, because owners contact NHTSA about notification letters and unavailable parts and those contacts are filed under the recall's own component label.

Check two: the magnitudes are window-dependent, so the ranking dies. Repeating the matched comparison at 18, 21, 24, 27 and 30 months, 19 of 28 models keep their sign but 9 do not - and the swings are large enough to invert a story:

Complaint arrival is lumpy and event-driven - a recall, a press cycle or an investigation lifts filings on the model years named. That is why this article names seven models and refuses to rank them. Anyone quoting a single percentage from this analysis for a single model is quoting the window, not the car.

Check three: recall correspondence. Excluding complaints whose narrative references a campaign number, recall notification, remedy, unavailable part or unreasonable delay leaves all seven stable-higher models in place and moves the 24-month count of rising models from 12 to 15 of 31. The finding is not recall mail.

All 39 pairs, with per-window deltas, the stability flag, the ex-correspondence column and the discarded lifetime column, are in the frozen dataset.

Does a rise mean the same defect continued?

No - and the Honda CR-V is in the seven specifically to make that limit concrete rather than leave it as a disclaimer. The CR-V's cutoff comes from 17V305000, a narrow recall of 2015-2016 cars that had received a replacement engine built with the wrong pistons. The MY2017 rise in engine complaints is the separate, well-documented 1.5-litre turbo oil-dilution problem. Two unrelated faults, one apparent continuity across a cutoff.

So the honest reading of a rise is: this model year is worth looking at. Establishing what is wrong means reading the complaint narratives and the manufacturer's filings - which is also why a component label on a recall lookup can describe one part while the complaints describe another.

What a VIN check can and cannot tell you here

If the car you are considering sits just outside a recall population, the sequence is short. Pull the VIN's written history to see what that specific car has been through and whether recalls are on record. Run the same VIN through NHTSA's own free tool for open-campaign status. Read the complaint narratives for your model year on NHTSA's site and see whether owners describe the failure the recall addressed. Then ask a franchised dealer to read the manufacturer's repair history against the VIN - that is the only place a completed remedy is recorded, which is the same reason an over-the-air fix can leave no service record at all.

A Zilocar report is $39.50 for a single report, or $29.50 for the first month and then $59.75 a month for 20 reports. There are also free sources worth using first, and NHTSA's recall lookup is one of them.

Method

Sources. Two NHTSA Office of Defects Investigation flat-file sets, both public domain, both retrieved 2026-11-07.

• Recalls: FLAT_RCL_POST_2010.zip from https://static.nhtsa.gov/odi/ffdd/rcl/. Archive 15,026,387 bytes; server Last-Modified 2026-11-07 08:03:54 GMT; member FLAT_RCL_POST_2010.txt 312,829,632 bytes, member timestamp 2026-11-07 08:02; testzip() clean. Tab-delimited, no header, 29 fields. 247,176 rows, of which 220,429 carry RCLTYPECD = 'V'.
• Complaints: COMPLAINTS_RECEIVED_2015-2019.zip (81,302,756 bytes), _2020-2024.zip (75,874,323 bytes) and _2025-2026.zip (43,507,472 bytes) from https://static.nhtsa.gov/odi/ffdd/cmpl/. 1,120,241 rows on a uniform 51-field layout on every row, of which 1,106,872 carry PROD_TYPE = 'V', resolving to 789,754 distinct ODINO. Field positions used: MAKETXT 4, MODELTXT 5, YEARTXT 6, COMPDESC 12, LDATE 17, CDESCR 20, PROD_TYPE 46.

Why not the combined complaint download. · FLAT_CMPL.zip was deliberately not used. That endpoint has repeatedly served a silently truncated extract under a fresh timestamp - on one occasion records stopping eight years early, on another a file 2.5% of its real size with a newest filing thirty years old - and each time the archive opened cleanly and parsed, so nothing downstream failed. The COMPLAINTS_RECEIVED_* period files are the current, complete equivalent.

Freshness assert, run before any computation. The newest LDATE across the three complaint files is 2026-11-04, three days before the run; the oldest is 2015-01-01. The run aborts if the newest filing is more than 30 days stale. This is checked on the newest record date, not on byte count and not on Last-Modified, because a stale-but-parseable file is the failure mode rather than a missing one.

Universe. A model qualifies if any recall with RCLTYPECD = 'V' has a component field matching the word ENGINE, giving 475 make/model combinations across 324 distinct engine campaigns once restricted to a 40-make allowlist of passenger-car and light-truck brands. The allowlist matters: 'V' means vehicle, not car, and the unrestricted basis carries motorcycles, heavy trucks, buses, motorhomes and towable industrial equipment. The cutoff is the maximum affected model year across that model's engine campaigns; the first uncovered model year is cutoff + 1.

Matched exposure, and N = 28. Counts are taken in a window of W months beginning 1 October of the preceding calendar year, approximating the on-sale date, so the two model years are observed at the same age. The analysis is run at W = 18, 21, 24, 27 and 30 months. A pair counts in a given window only if both windows close on or before the 2026-11-04 data end and both model years carry at least 30 complaints in window. 39 pairs are measurable in at least one window; 28 are measurable in at least three, and those 28 are the N for every stability claim. The window start is an approximation and launch dates vary by weeks, which is a further reason small differences are not meaningful.

Component definition. Engine complaints are identified on COMPDESC by exact set membership of {ENGINE, ENGINE AND ENGINE COOLING}. A prefix match is specifically avoided: NHTSA carries several top-level strings where one is a prefix of another, and merging them can decide a superlative. Both strings are kept because they are the same system under two intake vocabularies - ENGINE arrives overwhelmingly through the website and ENGINE AND ENGINE COOLING through the telephone hotline, so dropping either would rank an intake channel rather than a car.

Deduplication. Complaints are deduplicated on ODINO within each model-year cell, since the file carries one row per cited component and a row count would multiply a complaint citing several.

Results. Of the 28, 19 hold their sign in every window in which they are measurable: 7 consistently higher on the uncovered model year and 12 consistently lower. 9 are window-dependent.

Check one, which changed the answer. The intuitive lifetime-total comparison flips the sign on 9 of the 31 pairs measurable at 24 months. The Jeep Wrangler reads +25.9pp lifetime and −24.8pp matched; the Toyota Tundra −17.0pp lifetime and +1.2pp matched; the Jeep Grand Cherokee −14.2pp and +7.9pp. Two mechanisms drive it in the same direction: the uncovered year is the newer car with less road time, and a recall generates complaint traffic onto the years it covers. The lifetime column is published in the frozen CSV beside the matched ones, so the discarded version is auditable rather than merely asserted.

Check two, which killed the model-level ranking. Magnitudes move sharply with window length even where direction does not. The Kia Telluride spans +2.6 to +29.9pp and the Jeep Wrangler −24.8 to +29.3pp on nothing but W. Consequently no worst-model ranking is published anywhere in this article, the seven models are listed rather than ordered, and only direction is claimed. A single percentage for a single model from this analysis is a statement about the window.

Check three, recall correspondence. When an owner contacts NHTSA about a recall - usually because a repair part is unavailable - the contact is filed under that recall's own component label, with nothing in the structured fields to separate it from a fault report. Excluding complaints whose CDESCR references a campaign number, recall notification or notice, remedy, unavailable part or unreasonable repair delay leaves all seven stable-higher models in place and moves the 24-month count of rising models from 12 to 15 of 31. The leaders strengthen rather than weaken - the Telluride moves from +29.9 to +34.7 points at 24 months - so the finding is not recall mail.

Stated limit, not a result. The method detects that engine complaints rise past a cutoff, not that the same defect continues. The Honda CR-V is the worked counter-example and is published in the main table for that reason: its cutoff rests on 17V305000, a narrow recall of 2015-2016 cars that had received a replacement engine with incorrect pistons, while the MY2017 rise is the separate 1.5-litre turbo oil-dilution issue. Any single row is a question to investigate, not a defect established.

Corroboration. The named campaigns were independently re-queried against https://api.nhtsa.gov/recalls/campaignNumber: 24V381000 (102,092 units, Tundra MY2022-2023), 25V767000 (126,691 units, adding MY2024), 26V320000 (43,566 units, MY2024), 24V077000 (Telluride MY2024, valve springs that may break while driving), 25V274000 (GM 6.2L L87, 2021-2024) and 17V305000 (CR-V replacement engines). Model-year coverage and component strings matched the flat file in every case.

Known limits.

• A complaint count is not a defect rate. There is no denominator of cars on the road in this file, so a higher-selling model year will tend to generate more complaints. Matching exposure controls for age, not for volume. No claim here is a per-vehicle rate.
• The measurable set is small and selected. 28 pairs out of 475 candidate models. A pair is excluded whenever its windows fall outside the complaint files, which removes every cutoff older than about model year 2015 - including the Hyundai Sonata and Chevrolet Cruze that dominate a lifetime-basis ranking. This is a sample of recent cutoffs, not a census.
• Thin bases. The Ford F-250 Super Duty's uncovered year rests on 35 complaints and the Jeep Grand Cherokee's on 76. Direction is reported; the percentages carry real uncertainty.
• Sibling models disagree. The Silverado 1500 and the GMC Sierra 1500 share campaign 25V274000 and the same MY2024 cutoff, and read in opposite directions. Treat any single model as a lead, not a conclusion.
• Complaints are unverified reports submitted by owners, not confirmed failures, and attention effects are real: coverage of a recall raises filings on the models named.
• NHTSA republishes these files continuously, so a later download will not reproduce these counts exactly. The frozen CSV is the reproducible artefact.
• This analysis says nothing about whether any recall population will be expanded. That is a decision for the manufacturer and the regulator.

Frozen dataset. All 39 pairs, with per-window deltas at 18/21/24/27/30 months, the sign-stability flag and direction, the 24-month ex-correspondence delta, the discarded lifetime delta and the engine campaign count per model, are published at https://zilocar-news.github.io/zilocar-rss/data/recall-model-year-cutoff-engine-complaints-nhtsa-data-2026-2026-11-07.csv. Every number in this article can be recomputed from it, including the two versions that contradict it. Read the in_study_n28 column first: 11 of the 39 pairs are measurable in fewer than three windows and are marked no, carry insufficient_windows in place of a direction, and are excluded from the 28-pair N and from the group of seven. The 28 rows flagged yes reconcile exactly with this article: 7 higher on the uncovered year, 12 lower, 9 window-dependent.

Related checks

• Do recalls reach back to older used cars? - 7% of campaigns reach a decade-old model year and carry 28% of all recalled vehicles.
• How long does owner notification take? - why the letter may never have reached the car's current owner.
• Does an NHTSA investigation become a recall? - base rates for an open analysis like EA26005.

Sources

• NHTSA Office of Defects Investigation recall flat files, https://static.nhtsa.gov/odi/ffdd/rcl/ - FLAT_RCL_POST_2010.zip, retrieved 2026-11-07, Last-Modified 2026-11-07 08:03:54 GMT. Last verified 2026-11-07.
• NHTSA ODI complaint period files, https://static.nhtsa.gov/odi/ffdd/cmpl/ - COMPLAINTS_RECEIVED_2015-2019, _2020-2024 and _2025-2026, retrieved 2026-11-07, newest filing 2026-11-04. Last verified 2026-11-07.
• NHTSA recall API by campaign number, https://api.nhtsa.gov/recalls/campaignNumber - used to cross-check six campaigns. Last verified 2026-11-07.
• NHTSA VIN recall lookup, https://www.nhtsa.gov/recalls - the authority for open-recall status on one vehicle. Last verified 2026-11-07.
• NHTSA ODI complaint search, https://www.nhtsa.gov/vehicle-safety/complaints - owner-filed complaint narratives by model year. Last verified 2026-11-07.
• NMVTIS, https://vehiclehistory.gov - the authority for legal title brands. Last verified 2026-11-07.