PulsCar analysis of NHTSA public data, September 2026 Coolant leaks: 10,451 owner complaints, 34 recall campaigns, 1,142 owner-reported repair figures Licensed CC BY 4.0. Query run 18 September 2026; complaint set and every figure derived from it recomputed 18 September 2026 (see CORRECTION below). -------------------------------------------------------------------------- CORRECTION, 18 September 2026 -------------------------------------------------------------------------- The complaint set published earlier on this page (3,512) was built by matching the upper-case word lists below against the narrative WITHOUT case folding. NHTSA complaint narratives are upper case only through 2020; from 2021 on they are mixed case. The uppercase-only match therefore kept almost nothing received after 2020 (237 complaints in 2021, then 21, 18, 21, 12 and 8 in 2022-2026) and the page silently described a 2015-2020 file. Re-running the same rules case-insensitively returns 10,451 complaints, spread across every year, and every figure on this page and in the CSVs has been recomputed from that set. What changed materially: the fire share fell from 3.33% to 1.52% -- BELOW the 1.82% all-complaint baseline rather than nearly double it, which reverses the page's previous headline claim; the tow share rose from 3.73% to 5.03%, i.e. to the file's own level; the owner-reported repair figures grew from 454 to 1,142 and their median from $3,582 to $5,200. Two caveats about this correction, stated because they are real: 1. The rebuild re-implements the rules documented below from this file. Run case-sensitively, the rebuilt query returns 3,424 complaints against the 3,512 published earlier, so it is a faithful but not byte-identical reconstruction of the original script (2.5% tighter); the difference is in regex details that the original script, not this file, held. 2. Two phrase subsets could not be reconstructed at all -- the earlier "coolant and transmission fluid mixing" (448) and "coolant entering a cylinder" (156) figures came from patterns this file never spelled out. They have been replaced by two explicitly documented regexes (below), which are NOT the same measurement and are labelled differently on the page. The corrected set is within 0.7% of our companion page's independently documented coolant-loss set (10,523), which was always built case-insensitively -- that page is the cross-check for this one. -------------------------------------------------------------------------- SOURCE FILES -------------------------------------------------------------------------- NHTSA ODI complaint flat files COMPLAINTS_RECEIVED_2015-2019, _2020-2024, _2025-2026 -- https://static.nhtsa.gov/odi/ffdd/cmpl/ (dictionary CMPL.txt) 777,040 distinct vehicle complaints (ODINO, product type V) received 1 January 2015 to 8 September 2026. NHTSA recall flat file FLAT_RCL.zip. NHTSA defect investigation file FLAT_INV.zip (action number, make, model, model year, opened, closed, subject, summary). -------------------------------------------------------------------------- WHICH COMPLAINTS COUNT (the coolant-leak set: 10,451) -------------------------------------------------------------------------- A complaint counts when its narrative contains ONE SENTENCE in which a cooling-system word and a leak word appear within 70 characters of each other. ALL MATCHING IS CASE-INSENSITIVE (Python re.IGNORECASE); the word lists are printed here in capitals only for readability. cooling-system words: COOLANT | ANTIFREEZE | ANTI-FREEZE | RADIATOR | WATER PUMP | HEATER CORE | THERMOSTAT HOUSING | COOLANT RESERVOIR | EXPANSION TANK | OVERFLOW TANK | HEAD GASKET leak words: LEAK* | SEEP* | DRIP* | BURST* | RUPTURE* | CRACK* | SPRAY* | SPEW* | POOLING/POOLED UNDER | PUDDLE | LOSING COOLANT | LOSS OF COOLANT | COOLANT LOSS | LOW COOLANT | EMPTY EXCLUSIONS, applied per sentence: NO LEAK, NOT LEAK, DID NOT LEAK, DOES NOT LEAK, and leaks of a different fluid named in the same sentence (oil leak only, brake fluid leak, power steering fluid leak, transmission fluid leak, fuel or gas leak), plus cabin water leaks (water leak into the cabin or trunk, sunroof, windshield). NOTE on transmission fluid: a sentence that names ONLY a transmission fluid leak is excluded, but the large family of narratives describing a radiator's internal transmission cooler letting coolant and transmission fluid mix is INCLUDED, because those sentences name the radiator and a coolant leak. Makes excluded: motorcycle, heavy-truck, bus and RV manufacturers (the same list used in our other analyses), plus Cummins and FCCC. SAMPLE READING (required before publishing any keyword count) Thirty randomly drawn matches were read in full. Twenty-nine described a coolant leak, a coolant loss or a cooling-system component leaking. One did not (an airbag narrative that mentioned the radiator in passing). FALSE-POSITIVE RATE: 1 IN 30, ROUGHLY 3%. -------------------------------------------------------------------------- RESULTS -------------------------------------------------------------------------- Outcome flags (share answering Y, of 10,451): fire 1.52%, towed 5.03%, crash 0.63%, 74 injuries, 1 death. All-complaint baseline (777,040): fire 1.82%, towed 5.11%, crash 4.37%. The crash share is the finding: a coolant-leak complaint is about one seventh as likely as the average complaint to report a crash, tows at the file's own rate, and reports a fire slightly LESS often than the average complaint. (The earlier version of this page claimed the opposite about fire; see CORRECTION above.) Mileage (MILES field, 0 < miles < 400,000; 4,549 complaints report one): median 88,000; quartiles 62,000 and 116,000; 9.9% at or under 36,000 miles; 23.7% at or under 60,000; 62.6% at or under 100,000. All-complaint median for the same field: 70,000. Phrase subsets, share of the 10,451 narratives: radiator 1,637 (15.7%); head gasket 1,281 (12.3%); water pump 1,137 (10.9%); reservoir/overflow/expansion tank 920 (8.8%); thermostat 417 (4.0%); cracked block or head (including "porosity") 322 (3.1%); a named hose 288 (2.8%); oil or EGR cooler 144 (1.4%); heater core 111 (1.1%); water outlet or coolant housing 28 (0.3%); intake manifold gasket 11 (0.1%). Also: overheating 2,425 (23.2%); an engine replaced or needing replacement 1,504 (14.4%); mentions a recall 2,845 (27.2%); mentions a warranty 1,540 (14.7%); no fault found family 140 (1.3%). Two subsets are NEW measurements replacing figures that could not be reconstructed (see CORRECTION): names transmission fluid, a transmission cooler or cross-contamination -- regex (TRANSMISSION FLUID|TRANS FLUID|COOLANT (IN|INTO) THE TRANSMISSION|TRANSMISSION COOLER|CROSS.?CONTAMINAT|MILKSHAKE| STRAWBERRY) -- 276 (2.6%); names coolant intrusion, the combustion chamber, coolant in a cylinder or burning coolant -- regex (COOLANT INTRUSION|COMBUSTION CHAMBER| COOLANT (IN|INTO) THE CYLINDER|COOLANT (IN|INTO) THE ENGINE BLOCK| BURNING COOLANT) -- 850 (8.1%). Because owners name the part a shop told them about, these shares measure vocabulary, not cause frequency. Narratives can name more than one part. OWNER-REPORTED REPAIR FIGURES (1,142 figures) A dollar amount counts when: it appears in the narrative of a complaint in the coolant set; the surrounding +/-140 / +110 characters contain a cost word (repair, quote, quoted, estimate, cost, charged, paid, bill, fix, replace, rebuild) AND a cooling-system or engine word (coolant, antifreeze, radiator, water pump, head gasket, thermostat, heater core, reservoir, engine, transmission, hose, block, cylinder); the window does NOT contain a disqualifying word (for the car/vehicle/truck, purchased, bought, rental, deductible, down payment, trade-in, monthly, per month, a gallon, insurance, lawsuit, blue book, worth, tire, door, cluster, windshield, paint, brake pad, battery, lease, loan, sold it); and the value is between $50 and $20,000. Median $5,200; first quartile $2,000; third quartile $8,000. Bands: $50-$250 44 (3.9%); $251-$500 36 (3.2%); $501-$1,000 69 (6.0%); $1,001-$2,000 137 (12.0%); $2,001-$4,000 179 (15.7%); $4,001-$20,000 677 (59.3%). SAMPLE READING: twenty of the extracted figures were read with their context. Roughly three in twenty were not the price of the coolant repair itself -- an unrelated module, an owner quoting prices found in their own research rather than a quote received, or a repair to a different system in the same narrative. TREAT THE DISTRIBUTION AS CARRYING ROUGHLY ONE MIS-SORT IN SEVEN. THE LARGER CAVEAT, WHICH MATTERS MORE THAN THE REGEX: these are the prices of coolant leaks that became federal safety complaints. Nobody writes to the government about a $25 radiator cap. This distribution is evidence about the tail, NOT an estimate of what a coolant leak costs, and the article says so where it is used. RECALL CAMPAIGNS (34) Every campaign whose defect or consequence text names a coolant leak, a leaking radiator, a leaking or failing water pump, coolant intrusion, a loss of coolant, a leaking heater core or a coolant hose -- after excluding the motorcycle, bus and heavy-truck makers above. 34 distinct campaigns; 782,216 vehicles by the manufacturers' own potentially-affected counts (summed once per campaign). 22 of the 34 (65%) name FIRE in the consequence field. Campaign rows are published in coolant-leak-recall-campaigns.csv. INVESTIGATIONS Read in full and quoted in the article: DP12-004 (Nissan Pathfinder, Frontier, Xterra transmission-cooler petition, opened 11 June 2012, denied 22 October 2016); PE13-029 (2013 Pathfinder internal transmission oil cooler hose, opened 29 August 2013, closed 31 March 2014); PE18-007 (2013 Ford Escape 1.6L GTDI loss of motive power, opened 16 July 2018, closed 27 December 2024); DP26-008 (2015-2018 Ford Edge 2.0L EcoBoost coolant intrusion petition, opened 3 September 2026, still open at the time of writing). -------------------------------------------------------------------------- LIMITATIONS -- READ BEFORE QUOTING ANY NUMBER -------------------------------------------------------------------------- 1. Complaints are self-reported, unverified and voluntary. They are not failure rates and cannot be converted into one: NHTSA publishes no denominator. 2. Counts favour high-volume models and models whose problem has been publicized. Nothing here ranks reliability. 3. The complaint query carries roughly one mis-sort in thirty; the dollar extraction carries roughly one in seven. 4. Fire, towed, crash and mileage are the owner's own entries, and frightening events prompt filings, so these shares are inflated relative to the real world. The comparison that survives that bias is the one used in the article: fire versus the SAME field in the SAME file across all complaints. 5. "Porosity" is included in the cracked-block regex because Toyota's own recall 20V-064 uses that word for the same failure. -------------------------------------------------------------------------- FILES -------------------------------------------------------------------------- coolant-leak-by-make.csv make, complaints, miles reported, median miles, percent fire, percent towed, percent mentioning overheating. 8-complaint floor. coolant-leak-top-models.csv same columns plus top model year and its count, for the 60 most-reported models. coolant-leak-by-part-named.csv part named in the narrative, complaints, share of 10,451, percent fire, median miles. coolant-leak-owner-reported-costs.csv cost band, figures, share, plus median and quartiles. coolant-leak-recall-campaigns.csv campaign, manufacturer, report date, potentially affected, makes, whether the consequence text names fire. 34 rows. No personal data is published: no VIN, no city, no dealer name, no owner free text beyond short quoted snippets in the article itself.