PulsCar analysis of NHTSA public data, September 2026 Chevrolet Equinox: what 6,838 owner complaints, 26 recall campaigns and the 2.4L special-coverage bulletins say Licensed CC BY 4.0. Queries run 18 September 2026. -------------------------------------------------------------------------- SOURCE FILES -------------------------------------------------------------------------- NHTSA ODI complaint flat files COMPLAINTS_RECEIVED_2015-2019, _2020-2024, _2025-2026 https://static.nhtsa.gov/odi/ffdd/cmpl/ (field dictionary: CMPL.txt) 777,040 distinct vehicle complaints (ODINO, product type V) received 1 January 2015 to 8 September 2026. Of those, 6,838 name MAKE = CHEVROLET and MODEL = EQUINOX. Rows whose MODEL is EQUINOX EV (281 complaints) are a separate model in the flat file and are NOT included anywhere in this analysis. NHTSA recall flat file FLAT_RCL.zip (campaign, make, model, model year, component, potentially affected, report date, defect, consequence, remedy). Campaign details were re-read live on 18 September 2026 from https://api.nhtsa.gov/recalls/campaignNumber?campaignNumber= NHTSA manufacturer communication flat files TSBS_RECEIVED_2015-2019, _2020-2024, _2025-2026 https://static.nhtsa.gov/odi/ffdd/tsbs/ (field dictionary: TSBS.txt) 3,831 distinct NHTSA document IDs carry CHEVROLET / EQUINOX. NHTSA investigation flat file FLAT_INV.zip (action number, make, model, year, component, date opened, date closed, subject). Two special-coverage bulletins were opened as scanned PDFs and read in full: 14159D and 15285D, both dated April 2017. https://static.nhtsa.gov/odi/tsbs/2017/MC-10118904-9999.pdf (14159D) https://static.nhtsa.gov/odi/tsbs/2017/MC-10118905-9999.pdf (15285D) -------------------------------------------------------------------------- A MEASUREMENT PROBLEM THAT HAD TO BE FIXED FIRST -------------------------------------------------------------------------- NHTSA changed how it stores complaint narratives during 2021. Measured share of narratives that are entirely upper case, by year received: 2015-2020 about 100% 2023 1.5% 2021 44.1% 2024 1.0% 2022 2.1% 2025 1.0% 2026 0.8% A keyword regex written in upper case and matched case-sensitively therefore returns a plausible-looking number that silently excludes almost every complaint filed after mid-2021. Our first pass did exactly that and reported zero oil-consumption complaints for 2022 and 2023, which reading the narratives showed was false. EVERY query below is case-insensitive. -------------------------------------------------------------------------- WHICH COMPLAINTS COUNT AS OIL CONSUMPTION (812 of 6,838) -------------------------------------------------------------------------- A complaint enters the oil set when its narrative matches, case-insensitive: oil consumption | consumption of oil | consum\w* (a |one |1 )?quart | burn\w{0,3} (through |up )?(a |one |1 |\d )?quart | burning oil | burns oil | burnt oil | using (a |one |1 )?quart | uses (a |one |1 )?quart | add\w{0,3} (a |one |1 |\d )?quart | quart (of oil )?(every|per|each) | excessiv\w* (engine |amount of )?oil | oil (usage|burning) | piston ring Deliberately NOT included: "low oil", "oil level", "oil light", "oil leak". An earlier draft included the first two and pulled in oil-pressure-warning and oil-leak narratives, which are different complaints; dropping them moved the set from 1,000-plus matches to 812. SAMPLE READING (required before publishing any keyword count) Thirty matches were drawn at random (seed 20260918) and read in full. All 30 were genuine reports of an engine using oil between changes. Zero mis-sorts in 30 puts the upper bound on the false-positive rate at roughly 10% at 95% confidence (the rule of three); it does NOT mean the rate is zero. The query is narrow because the phrases are specific. What it will miss: an owner who describes the consequence (a seized engine, a knock) without ever saying the engine used oil. The set is therefore a floor, not a census. -------------------------------------------------------------------------- MILEAGE AND RATE FIELDS -------------------------------------------------------------------------- Mileage: NHTSA's MILES field, rows with 0 < miles < 400,000 only. Oil set: 446 of 812 report a usable mileage. Median 97,250; first quartile 75,000; third quartile 116,000. 47.5% at or above 100,000 miles; 11.4% at or under 60,000; 3.6% at or under 36,000. All Equinox complaints: 3,972 usable mileages, median 82,000. All-complaint baseline (whole file): 392,935 usable mileages, median 70,000, 26.9% at or under 36,000. Towed / crash / fire: NHTSA's TOWED, CRASH and FIRE flags, share = Y. Oil set: towed 2.59%, crash 0.74%, fire 0.74%, 8 injuries, 0 deaths. All Equinox: towed 5.82%, crash 3.77%, fire 1.04%. All-complaint baseline: towed 5.11%, crash 4.37%, fire 1.82%. Owner-reported dollar figures: every $N in the oil-set narratives with 100 <= N <= 30,000. 172 figures inside 124 of the 812 complaints. Median $2,950; first quartile $1,400; third quartile $5,800; largest $15,000. 65.7% are $2,000 or more. THIS IS NOT A PRICE. It is the expensive tail: an owner files a federal complaint over a four-figure engine bill, not over a $90 top-up, and one complaint can contribute several figures (a quote, a paid invoice, a trade-in loss). -------------------------------------------------------------------------- THE CHAIN COMPARISON (chain-phrases-oil-vs-other.csv) -------------------------------------------------------------------------- The consolidated class action described a sequence: piston ring wear -> excessive oil consumption -> knocking, low oil pressure, fouled plugs, timing chain damage, engine failure. To test whether owners describe that sequence, each phrase family was counted in the 812-complaint oil set and in the other 6,026 Equinox complaints, and the two shares divided. WHAT THE RATIO IS AND IS NOT. It is co-occurrence inside self-reported narratives. It is not causation, not a failure rate, and not a comparison between vehicles. Both sets are self-selected, and an owner who already believes the engine is burning oil is more likely to mention a knock in the same sentence. The comparison is still worth publishing because the control group is the SAME vehicle and the same filing system. Two phrase families were excluded from the ratio because the selection query implies them: "piston ring" is inside the oil regex, so counting "piston" against the other set is circular. "Documented oil consumption test" is published with its within-set share only (80 of 812, 9.9%) and no ratio, for the same reason. -------------------------------------------------------------------------- RECALLS (recall-campaigns.csv) -------------------------------------------------------------------------- 26 distinct campaigns name the Chevrolet Equinox in the flat file, covering 2,778,331 potentially affected vehicles (summed once per campaign, not once per model year row). Not one campaign is for engine oil consumption. Two campaigns mention oil at all, and neither is engine oil consumption: 18V-576 (rear brake caliper piston coating) and 20V-668 (start-stop accumulator endcap bolts, transmission oil leak). The potentially-affected count is the whole campaign population across every make and model in it, not the Equinox share; NHTSA does not publish a per-model split. Movement of complaint counts around a campaign was computed by counting, for a given model year and keyword, the complaints received in each calendar year. This is descriptive only. A recall changes the number of complaints for at least three reasons that cannot be separated here: the remedy works, owners stop filing because the problem is now official, and publicity briefly makes owners file more. -------------------------------------------------------------------------- GENERATION COMPARISON (themes-by-generation.csv) -------------------------------------------------------------------------- Model years 2010-2017 (4,362 complaints) against 2018-2026 (1,724). Each theme is the share of complaints whose full narrative matches a plain keyword family; themes overlap and do not sum to 100. Model years 2005-2009 (752 complaints) are the first generation and are excluded from both columns. LIMITATION THAT MATTERS MOST HERE: the two columns are not comparable as risk. A 2013 vehicle has had up to eleven years to accumulate complaints and is out of warranty; a 2025 vehicle has had months and is inside it. Newer vehicles are over-represented in early-life themes (brakes, software, infotainment) and under-represented in wear themes. Nothing in this table says one generation is more or less reliable than the other. -------------------------------------------------------------------------- BULLETINS READ -------------------------------------------------------------------------- Where the article quotes a bulletin word for word, the words come either from a scanned PDF we opened (14159D, 15285D) or from NHTSA's own summary record of the manufacturer communication, and the article says which. NHTSA's summary is its paraphrase of the manufacturer's document, not always the manufacturer's wording. Documents cited by NHTSA document ID, all read in the manufacturer communication file on 18 September 2026: 10118904 14159D April 2017 special coverage, 2010, PDF opened 10118905 15285D April 2017 special coverage, 2011, PDF opened 10118906 16118 October 2017 special coverage, 2012 10171429 16118 01 18 Feb 2020 revised service procedure, 2012 10171426 N192291100 18 Feb 2020 special coverage, 2013 (2.4L RPO LEA) 10171430 N192291100 01 18 Feb 2020 same, revised warranty 10183216 N192287100 30 Nov 2020 converters and O2 sensors, 2012-2013 10090101 14882 1 Feb 2017 PCV freeze after high oil consumption 10057696 14882 22 Jun 2015 owner notification letter, 2010-2014 10116880 13-06-01-003H 31 Aug 2017 piston, rod and bearing procedure 10112116 PIP5163A 15 Aug 2017 cylinder bore damage as a cause 10114867 PIP5331 15 Aug 2017 rattle, tick or knock diagnosis 10137634 16-NA-367 21 May 2018 dexos1 oil on LFV and LE2 engines 10119010 GCUS-3-702 15 Oct 2017 emission recall 43180, HPFP wear The scanned PDF URL for a manufacturer communication is not derivable from the document ID; only the 2017 GCUS batch resolved to a static.nhtsa.gov path. The rest are cited by document ID, which is what the public Manufacturer Communications search takes as input. -------------------------------------------------------------------------- INVESTIGATIONS -------------------------------------------------------------------------- RQ18-002 opened 2 Nov 2018, closed 19 Dec 2019, "Windshield Wiper Transmission Failure", Equinox model years 2010-2012 and 2014-2016. EA19-005 opened 13 Dec 2019, closed 28 Nov 2022, "Front Wiper Transmission Inoperative", same model years. Recall 22V-165 carries EA19005 as its NHTSA action number. PE20-009 opened 1 Jun 2020, closed 3 Jun 2022, "Momentary Increased Steering Effort", Equinox 2010-2012. EA26-006 opened 21 Aug 2026, open, "Electronic Brake Module Component Failure", EQUINOX EV 2024-2026 — a different model in the flat file, listed here only so readers do not mistake it for this one. -------------------------------------------------------------------------- LIMITATIONS -- READ BEFORE QUOTING ANY NUMBER -------------------------------------------------------------------------- 1. Complaints are self-reported, unverified and voluntary. They are not failure rates and cannot be turned into one, because NHTSA publishes no denominator: nobody knows how many 2013 Equinoxes are still on the road. 2. Counts follow attention as much as mechanics. A defect that has been written about, litigated or recalled collects complaints; one that has not, does not. Nothing here says the Equinox is more or less reliable than any other vehicle, and no honest reading of complaint counts can say that. 3. Shares by filing year are published instead of raw counts wherever the denominator is shrinking, because the 2010-2013 fleet leaves the road year by year: 188 complaints about those model years in 2015, 23 in 2026. 4. Mileage, towed, crash and fire are the owner's own entries. Towed and crash shares in any complaint set are inflated relative to the real world, because a frightening event is what prompts a filing. 5. The engine option is not in the complaint file. NHTSA records the model, not the RPO code, so a complaint about a V6 Equinox and a complaint about a 2.4L Equinox are indistinguishable by field. The oil set is separated by what the owner wrote, not by what engine the car has. 6. Model year in the complaint file is the owner's entry and is sometimes wrong. -------------------------------------------------------------------------- FILES -------------------------------------------------------------------------- oil-complaints-by-model-year.csv model_year, equinox_complaints, oil_consumption_complaints, pct_of_that_year, mileages_reported, median_miles, pct_at_or_over_100000_miles. oil-complaints-by-filing-year.csv year_complaint_received, my2010_2013_equinox_complaints, of_which_oil_consumption, pct, oil_consumption_complaints_all_model_years. chain-phrases-oil-vs-other.csv phrase_family, oil_set_812_n, oil_set_pct, other_equinox_6026_n, other_equinox_pct, ratio. recall-campaigns.csv campaign, report_date, component, potentially_affected, equinox_model_years, consequence. themes-by-generation.csv theme, my2010_2017_n, my2010_2017_pct_of_4362, my2018_2026_n, my2018_2026_pct_of_1724. No personal data is published: no VIN, no city, no dealer name, no owner free text beyond short quoted snippets in the article itself.