Get a Full Vehicle History Report

We Counted Every Fuel-System Complaint NHTSA Has Logged Since 2015. Most of Them Are Recall Paperwork, Not Fuel Faults

· Zilocar Editorial

Ford's 223,472-truck fuel tank recall put fuel systems in the news this week, so we counted every gasoline fuel-system complaint in NHTSA's own files: 7,308 since 2015, out of 777,040 vehicle complaints. Then we read the text of each one. 58.9% reference a recall rather than report a fault, and removing them reverses every leaderboard the raw counts produce.

Users who trust ZILOCAR
Trusted by thousands of users all over the globe

Key facts

  • 7,308 complaints cite a FUEL SYSTEM, GASOLINE component, out of 777,040 distinct vehicle complaints filed between 2015-01-01 and 2026-09-08. That is 0.94% of all complaints.
  • 4,308 of them (58.9%) reference a recall — an NHTSA campaign number, a recall notification, a remedy, or a part that was unavailable — rather than reporting an independent fault.
  • A single recall accounts for 1,301 citations, 17.8% of the whole bucket: Honda's 23V858000 in-tank fuel pump recall, 2,490,460 vehicles across roughly 20 Honda and Acura nameplates.
  • Both leaderboards reverse when recall correspondence is removed. On volume the Honda CR-V falls from 1st (429) to 3rd (82) — 80.9% of its fuel-system complaints were recall traffic. On share the Ford Bronco Sport falls from 1st at 11.00% out of the top five entirely, because 92.5% of its were.
  • The reversal holds under both definitions we tested, strict and loose. Which model replaces the leader does not hold, so this page names no new "worst" model.
  • Recall-referencing complaints peaked at 1,393 in 2024 while fault-describing complaints stayed flat at 474 — a recall-shaped spike, not a defect-shaped one.

Why do most fuel-system complaints turn out to be about recalls?

Because NHTSA files a complaint under the component label of the recall it concerns. When an owner contacts the agency about a recall — overwhelmingly because the repair part is not available — that correspondence is deposited in the same bucket as a genuine fault report, and nothing in the structured fields separates them. Only the free-text description does.

The pattern is unmistakable once read. A representative Bronco Sport entry:

"The contact received notification of NHTSA Campaign Number: 22V859000 (Fuel System, Gasoline) however, the part to do the recall repair was not yet available. The contact stated that the manufacturer had exceeded a reasonable amount of time for the repair."

That is a real and serious consumer problem — it is just not evidence that the model has a fuel defect. It is evidence that a recall remedy was slow.

Which used cars have the most fuel-system complaints?

As filed, the Honda CR-V leads on volume and the Ford Bronco Sport on share. Neither survives de-confounding. The table gives both the raw figure and the figure after recall-referencing complaints are removed, so the gap between them is visible per model.

Make / modelTotal complaintsFuel-systemShareReference a recallAfter removalShare after
Honda CR-V12,6084293.40%347 (80.9%)820.65%
Ford Escape19,5823801.94%274 (72.1%)1060.54%
Ford Focus12,0873653.02%277 (75.9%)880.73%
Honda Accord11,3363012.66%252 (83.7%)490.43%
Honda Pilot7,3082042.79%165 (80.9%)390.53%
Honda Civic8,2191832.23%165 (90.2%)180.22%
Hyundai Sonata11,5441811.57%137 (75.7%)440.38%
Ford F-15021,2961750.82%100 (57.1%)750.35%
Ford Bronco Sport1,33614711.00%136 (92.5%)110.82%
Honda Odyssey6,0781201.97%98 (81.7%)220.36%
Ram 150011,4721161.01%59 (50.9%)570.50%
Jeep Grand Cherokee14,1431050.74%50 (47.6%)550.39%
Ford Fusion18,0111000.56%35 (35.0%)650.36%

The column that matters is the fourth. It ranges from 35.0% on the Ford Fusion to 92.5% on the Bronco Sport, which is why the two rankings disagree so violently: the raw count is measuring recall exposure as much as anything mechanical.

The Bronco Sport is the extreme case and its cause is documented. Its most-cited campaign is 25V467, the fuel-injector recall we covered in July, which followed an earlier remedy that did not hold. Owners wrote to NHTSA in volume about that repair, and on a nameplate with only 1,336 total complaints the correspondence alone is enough to produce an 11.00% "fuel-system share" out of 147 complaints, of which just 11 describe a fault once recall references are removed.

Does the Honda result mean Hondas have a fuel problem?

No, and this is the clearest single illustration in the data. Honda and Acura nameplates account for 1,714 of the 7,308 fuel-system complaints, 23.5% of the bucket. Of those, 84.7% reference a recall, leaving 263. Their share of the de-confounded total falls to 8.8%.

The cause is one campaign. 23V858000, filed 2023-12-18, recalls about 2,490,460 vehicles because the in-tank fuel pump may fail and stall the engine — across the Accord, Civic, CR-V, HR-V, Ridgeline, Odyssey, Pilot, Passport, Fit, Insight, Clarity and the Acura ILX, MDX, RDX, RLX, TLX and NSX. Every one of those nameplates that appears high on the as-filed share board appears there because of it: the HR-V at 4.26% is 96.2% recall-referencing, the Acura TLX at 3.21% is 89.3%, the Passport at 3.58% is 88.9%.

A recall that large generates correspondence at a scale no ordinary defect does. Counting it as defect signal manufactures a brand-level finding out of a single remedy backlog.

What did the robustness checks do to the finding?

We ran the check most likely to reverse our own headline, and report it either way. "References a recall" is a loose test, because 1,261 complaints do both — they cite a campaign and describe a fuel symptom. So we re-ran everything under a stricter definition that removes only complaints referencing a recall and describing no fault: 3,047 complaints, 41.7%.

LeaderboardAs filedExcluding all recall-referencingExcluding strict recall-only
Volume, 1stHonda CR-V, 429Ford Escape, 106Ford Focus, 183
Volume, 2ndFord Escape, 380Ford Focus, 88Ford Escape, 159
Volume, 3rdFord Focus, 365Honda CR-V, 82Honda CR-V, 130
Share, 1stFord Bronco Sport, 11.00%Chevrolet HHR, 3.47%Land Rover Range Rover Sport, 5.18%
Share, 2ndLand Rover Range Rover Sport, 6.22%Saturn Ion, 2.98%Saturn Ion, 3.76%

What survives every cut: the as-filed leaders are artifacts. The Honda CR-V does not lead on volume and the Ford Bronco Sport does not lead on share under either de-confounded definition.

What does not survive: the identity of the replacement. Volume leadership moves to the Escape under one definition and the Focus under the other; share leadership moves to the Chevrolet HHR under one and the Range Rover Sport under the other. We therefore name no new worst model, and any quotation of this work should not either.

One further limit we measured rather than assumed: the FUEL SYSTEM, GASOLINE label is itself an upper bound on fuel faults. Only 45.2% of the bucket's text describes a recognisable fuel symptom under our narrow screen, though 92.5% mentions fuel or petrol at all. So the component label over-counts in both directions — recall traffic inflates it, and loose labelling blurs it.

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

A VIN check is the right tool for the recall half of this story and the wrong tool for the complaint half.

It can tell you whether a specific used vehicle sits inside an open safety recall and how many are attached, including fuel-system campaigns such as 26V578000 or 23V858000. Alongside that, a Zilocar report adds the car's accident and damage records including airbag deployment, odometer readings and rollback indications, theft records, junk and salvage auction history, ownership history, sales-listing history, specs, NHTSA and IIHS ratings and a valuation.

It cannot tell you whether the recall repair was actually performed — that is the precise gap these complaints are about, and it comes only from a franchise dealer or the manufacturer's owner portal using the same VIN. It also cannot show NHTSA investigations that have not become recalls, per-unit firmware or software state, or a legal title-brand classification.

And no VIN report diagnoses a fuel leak. Complaints are model-level, self-reported and unverified; the car in front of you needs a physical inspection of the tank, lines and filler by someone who can put it on a lift.

By the numbers

  • 777,040 distinct vehicle complaints, filed 2015-01-01 to 2026-09-08
  • 7,308 citing a gasoline fuel-system component (0.94%)
  • 4,308 referencing a recall (58.9%); 3,047 referencing one and describing no fault (41.7%)
  • 3,000 remaining under the loose cut; 4,261 under the strict cut
  • 1,301 citations of a single campaign, 23V858000 (17.8% of the bucket)
  • 51,740 complaints citing FUEL/PROPULSION SYSTEM, the electric and hybrid propulsion label, deliberately excluded; 1,264 citing FUEL SYSTEM, DIESEL, also excluded
  • 656 make/model labels carry at least one fuel-system complaint
  • 117 of the 7,308 carry the owner-entered fire flag, on four labels with five or more; published as a floor, never ranked

What is the Ford F-150 recall that prompted this?

NHTSA campaign 26V578000, Ford's internal 26S69, received 2026-09-09, covers 223,472 model-year 2023-2027 Ford F-150 trucks. The fuel tank may be improperly secured, and can leak fuel or detach from the vehicle while driving; NHTSA's stated consequence is that a detached tank creates a road hazard and a fuel leak can cause an engine stall or raise fire risk. Dealers will inspect and replace the fuel tank straps free of charge. Owner notification letters were expected 2026-09-21, and VINs became searchable on NHTSA.gov from 2026-09-14.

Two things to be accurate about: the campaign carries no do-not-drive and no park-outside orderparkIt and parkOutSide both return false from NHTSA's API — and it is a 2023-2027 truck, so it sits at the newest edge of the used market rather than in its middle.

On present evidence this recall will do to the F-150's complaint count what 23V858 did to Honda's. The F-150 already sits at 57.1% recall-referencing, the lowest of the high-volume Fords, and that figure should be expected to climb through 2027 without telling anyone anything new about the truck.

Frequently asked questions

Method

Source files. NHTSA Office of Defects Investigation consumer-complaint period files, not the combined download: COMPLAINTS_RECEIVED_2015-2019.zip (81,302,701 bytes), COMPLAINTS_RECEIVED_2020-2024.zip (75,873,826 bytes) and COMPLAINTS_RECEIVED_2025-2026.zip (38,983,619 bytes), all retrieved and processed 2026-09-13. Tab-delimited, 51 fields, no header row, layout per CMPL.txt.

Freshness was asserted before anything was computed, and this is why the period files were used. On the morning of this run NHTSA's combined FLAT_CMPL.zip was serving 56,997 rows whose newest filing was 1996-10-10 — 3,225,032 bytes against 371,393,017 three days earlier — with the archive's CRC intact and a Last-Modified of 2026-09-11, so nothing about the download looked wrong. We re-fetched it to a clean path with byte accounting and reproduced the same file, while the period files from the same directory arrived complete, which rules out our own transport. The period files' newest filing is 2026-09-08, five days before the run. Any figure on this page would have been wrong by a factor of forty had the combined file been trusted.

Universe. Keep PROD_TYPE = 'V' (vehicles, dropping 13,016 equipment, tyre and child-seat rows) and LDATE >= 2015-01-01. Rows are one per cited component, so all 1,100,893 rows were deduplicated to distinct ODINO complaint ids, giving N = 777,040 complaints across 656 make/model labels carrying a fuel-system complaint. Every figure on this page is a count of complaints, never of rows, vehicles, owners or confirmed faults. No duplicate CMPLID appears across the three files.

The fuel-system cut. A complaint qualifies when any of its component labels (COMPDESC) begins FUEL SYSTEM, GASOLINE, which captures the bare label (84.9% of rows) and all sub-nodes — DELIVERY:FUEL PUMP, STORAGE:TANK ASSEMBLY, FUEL INJECTION SYSTEM:INJECTORS, STORAGE:EVAPORATIVE EMISSIONS:CANISTER and the rest. FUEL/PROPULSION SYSTEM (51,740 complaints, the electric and hybrid propulsion label) and FUEL SYSTEM, DIESEL (1,264) are excluded and reported separately, so the gasoline fuel-leak framing stays honest.

The classification, which is the load-bearing step. Each qualifying complaint's free-text CDESCR was screened twice. Recall-referencing: mentions an NHTSA campaign number, a recall notification, notice, remedy or repair, or a part that was unavailable or exceeded a reasonable time. Fault-describing: names a fuel leak, smell or odour, fuel pump, tank, injector, rail, line, filler or cap, fuel gauge or sender, fuel starvation, evaporative canister, or fuel spilling. The two are not exclusive: 4,308 are recall-referencing, 3,349 are fault-describing, 1,261 are both, and 912 (12.5%) match neither.

Both definitions are published because the choice changes the answer. The loose cut removes all 4,308 recall-referencing complaints, leaving 3,000. The strict cut removes only the 3,047 that reference a recall and describe no fault, leaving 4,261. The headline claim — that the as-filed leaders are artifacts — is the one that survives both.

Ranking rules. Raw ranking is by complaint count. Share ranking is each model's fuel-system complaints over its own total complaints, with a 500-complaint floor to suppress low-N noise; without it, bus and specialty makers with a few dozen complaints dominate. Model labels were swept for consolidation, because NHTSA's MODELTXT is free text and one nameplate can occupy several strings: every same-make pair where one label is a prefix of another was tested for model-year overlap. All pairs above a 40-complaint floor — BRONCO/BRONCO SPORT, CR-V/CR-V HYBRID, ACCORD/ACCORD HYBRID, SILVERADO/SILVERADO 1500, F-150/F-150 HYBRID and twenty more — showed overlapping model years, i.e. genuinely different vehicles, so none were merged; merging the CR-V with the CR-V Hybrid moves its share from 3.40% to 3.45% and changes no ranking.

Known limits. (1) Complaints are consumer-reported and unverified — NHTSA does not adjudicate them, and none is a confirmed defect. (2) There is no denominator: NHTSA publishes no sales or registration counts, so nothing here is a failure rate, and the share metric controls for a model's complaint volume but not for its popularity or its publicity. (3) The keyword screens are approximate in both directions — a fault described in unusual words is missed, and a passing mention of a recall flags a complaint that may still be a genuine report; that is exactly why both definitions are published. (4) Only 45.2% of the bucket's text describes a recognisable fuel symptom, so the component label is an upper bound on fuel faults independently of the recall issue. (5) The consumer-entered FIRE, INJURED and DEATHS fields are floors, not measurements, and are not ranked here. (6) Model-level only: nothing on this page describes an individual vehicle. (7) NHTSA republishes these files continuously, so a re-pull will not reproduce these counts exactly — the frozen CSV is the reproducible artefact.

Sources

Last verified: 2026-09-13