PulsCar analysis of NHTSA public data, September 2026 Engine oil leaks: 4,197 owner complaints, 107 recall campaigns, 1,290 manufacturer communications 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/ (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 -- one row per make/model/model year per campaign; campaigns counted as DISTINCT campaign numbers. NHTSA manufacturer communications (TSB) flat file FLAT_TSBS.zip. Part 573 Safety Recall Reports and manufacturer communications downloaded from static.nhtsa.gov, listed at the bottom of this file. -------------------------------------------------------------------------- WHICH COMPLAINTS COUNT (the oil-leak set: 4,197) -------------------------------------------------------------------------- STEP 1 -- full-text search of the complaint narrative (FTS5) for any of: "oil leak" | "oil leaks" | "oil leaking" | "leaking oil" | "leaks oil" 5,009 complaints matched. STEP 2 -- three exclusions, applied to every occurrence of the phrase: (a) NEGATION. An occurrence does not count if one of no | not | never | without | n't | nor | neither | zero appears in the 60 characters before it and no sentence boundary separates them. This removes the very large family of narratives whose subject is oil DISAPPEARING with no leak to show for it -- "no visible oil leak, the oil is going somewhere" -- which is oil consumption, a different problem with a different repair. 707 complaints had no surviving occurrence and were dropped. (b) OIL THAT IS NOT ENGINE OIL. An occurrence does not count if its 140-character window names a differential, transfer case, PTO, gear oil, front or rear axle, gearbox, hydraulic circuit, power steering or transmission fluid AND does not also name the engine, a valve cover, an oil pan, a filter, a gasket or a seal. 91 complaints were dropped for naming only non-engine oil. (c) NOT A PASSENGER VEHICLE. 14 complaints on heavy-truck, bus, RV and motorcycle makes were dropped (Freightliner, Kenworth, Peterbilt, Mack, International, Western Star, Hino, Blue Bird, Thomas Built, Winnebago, Thor, Jayco, Keystone, Newmar, Tiffin, Coachmen, Grand Design, Entegra, Dutchmen, Prime Time, Crossroads, Spartan, Gillig, New Flyer, Autocar, Starcraft, Alliance, Harley-Davidson, Forest River, Heartland). 4,197 complaints survive. This is the denominator for every percentage below unless another one is named. SAMPLE READING (required before publishing any keyword count) BEFORE the filter: 40 randomly drawn raw matches were read in full. Six were negations ("no visible oil leak"), two named non-engine oil (a rear differential and a PTO unit). FALSE-POSITIVE RATE BEFORE FILTERING: 8 IN 40, 20%. That is what motivated exclusions (a) and (b). AFTER the filter: 30 randomly drawn surviving matches were read in full. All 30 described an engine oil leak on the complainant's own vehicle. One was borderline -- a brake vacuum pump leaking oil, which is engine oil but not a gasket or seal in the sense this article uses. RESIDUAL FALSE-POSITIVE RATE: AT MOST 1 IN 30, ROUGHLY 3%. -------------------------------------------------------------------------- WHICH ADDRESS A COMPLAINT NAMES -------------------------------------------------------------------------- A complaint is counted under an address when its narrative contains one of that address's phrases anywhere. Complaints naming two addresses are counted under both, so the address column does not sum to 4,197. Oil filter housing / oil cooler : filter housing | oil cooler | filter adapter | filter adaptor | oil filter assembly Oil pan or drain plug : oil pan | oil sump | drain plug | drain bolt Valve cover gasket : valve cover | cam cover | rocker cover | cylinder head cover Cylinder head / head gasket : head gasket | cylinder head | cracked head Timing cover / front crank seal : timing cover | timing chain cover | front cover | crankshaft seal | crank seal | front main seal | harmonic balancer seal Rear main seal : rear main PCV / crankcase ventilation : PCV | crankcase vent Turbocharger oil line : turbo (within 25 characters of) oil, line or hose | oil feed/supply/return line Oil pressure switch or sensor : oil pressure switch | sensor | sending unit | sender Oil line, hose or tube : oil line | oil hose | oil tube | oil pipe Camshaft seal / VVT solenoid : cam seal | camshaft seal | VVT solenoid | VVTi | variable valve timing solenoid 1,751 of 4,197 (41.7%) name at least one address. The other 58.3% report a leak without saying where it was found, which is itself the point of the article: most owners arrive without an address. -------------------------------------------------------------------------- RESULTS -------------------------------------------------------------------------- Outcome flags (share answering Y, of 4,197): fire 4.10%, towed 6.10%, crash 0.86%, 34 injuries, 0 deaths. All-complaint baseline (777,040): fire 1.82%, towed 5.11%, crash 4.37%. THE FINDING: an oil-leak complaint is about 2.3 times as likely as the average complaint to report a fire and one fifth as likely to report a crash. Oil leaks are a burn statistic, not a collision one. Mileage (MILES field, 0 < miles < 400,000; 2,173 complaints report one): median 81,000; quartiles 52,000 and 114,000. All-complaint baseline median: 70,000. Fire rate by address (same set; see oil-leak-complaints-by-address.csv): valve cover 5.21%, cylinder head 3.83%, filter housing/cooler 3.10%, oil pan/drain plug 1.81%, PCV 0.78%, and ZERO reported fires across the 466 complaints naming a timing cover/front seal, rear main seal, turbo oil line, oil pressure switch, oil line or camshaft seal. LIMITATION: these are counts of self-reported outcomes inside a self-selected set, not fire rates per vehicle on the road. What the gradient supports is a statement about GEOGRAPHY -- the addresses that sit above an exhaust manifold are the ones whose complaints mention fire -- not a ranking of which part is most dangerous. Language subsets (of 4,197): 453 (10.79%) use hot-exhaust, burning-smell, smoke or fire-hazard language. Within that subset the fire flag is 13.25%, against 4.10% for the set. 139 (3.31%) describe engine seizure, engine failure or an oil-pressure warning lamp. 57 (1.36%) say a shop could not find, duplicate or locate the leak. 598 (14.25%) mention a warranty. Tow rate by address: the rear-main-seal subset is the highest at 8.82% against 6.10% for the set and 5.11% for all complaints. Model-year concentration (oil-leak-top-models.csv): Jeep Wrangler is the most-complained nameplate in the set at 206 complaints. 143 of those 206 (69.4%) name the oil filter housing, cooler or filter adapter, and 127 of the 206 are model year 2014 alone. Jeep Grand Cherokee 62.0% and Jeep Cherokee 51.3% name the same part. Hyundai Veracruz: 34.0% name a valve cover (recall 14V-415 covers exactly that part on that vehicle). Ford F-150: 43.3% name the oil pan or drain plug, the highest pan share of any model in the top 45. Recalls (oil-leak-recall-campaigns.csv): 107 distinct campaigns 2015-2026 whose Part 573 defect or consequence text names an oil leak; 2,423,121 vehicles in the sum of campaign populations (campaign populations overlap across makes and are not a vehicle count). 71 of the 107 (66.4%) say FIRE in the defect or consequence text. Manufacturer communications: Full-text search of the TSB file for "oil leak" | "oil leaks" | "leaking oil" | "oil seepage" returns 1,152 distinct NHTSA document ids, 1,290 distinct make + bulletin-number pairs. Largest by make: Chevrolet 118, Honda 105, Volkswagen 85, GMC 79, Buick 69, Ford 66. -------------------------------------------------------------------------- LIMITATIONS -------------------------------------------------------------------------- 1. Complaints are self-reported, unverified and not normalized by how many vehicles are on the road. A make with more complaints may simply have sold more cars or have more motivated owners. Nothing here says which vehicle leaks more often, and no such claim is made in the article. 2. The address counts measure WHAT OWNERS WROTE, not what a technician confirmed. Many narratives repeat a shop's verdict; some repeat a forum's. The Wrangler filter-housing share is strong enough to survive that caveat; a 15-complaint address is not. 3. Fire, tow and crash are structured Y/N fields filled in at intake. The article notes where the structured field and the narrative disagree. 4. Recall campaign populations are "potentially affected" counts filed by the manufacturer, summed across campaigns that can cover the same car. -------------------------------------------------------------------------- PRIMARY DOCUMENTS OPENED FOR THIS ARTICLE (18 September 2026) -------------------------------------------------------------------------- Ford SSM 47420, 10 September 2018 (NHTSA 10144211) https://static.nhtsa.gov/odi/tsbs/2018/MC-10144211-9999.pdf Ford SSM 47265, 21 May 2018 (NHTSA 10137145) https://static.nhtsa.gov/odi/tsbs/2018/MC-10137145-9999.pdf Stellantis STAR case S2009000004, version 4.3 dated 28 October 2020 (NHTSA 10184415) https://static.nhtsa.gov/odi/tsbs/2020/MC-10184415-9999.pdf FCA GPOP tech tip 68105583A, 15 June 2016 (NHTSA 10139934) https://static.nhtsa.gov/odi/tsbs/2016/MC-10139934-9999.pdf FCA GPOP tech tip 9003357, 21 June 2017 (NHTSA 10138863) https://static.nhtsa.gov/odi/tsbs/2017/MC-10138863-9999.pdf GM PIP3343E, 15 August 2017 (NHTSA 10111783) https://static.nhtsa.gov/odi/tsbs/2014/MC-10111783-9999.pdf Recall 14V-415 (Hyundai Veracruz valve cover / alternator) Recall 22V-191 (Ford Escape and Bronco Sport oil separator housing) Recall 17V-209 (Ford 1.6L GTDI cracked cylinder head) Recall 19V-063 (Hyundai Tucson oil pan) all four read via https://api.nhtsa.gov/recalls/campaignNumber?campaignNumber= ------------------------------------------------------------------------------ CASE SENSITIVITY — SECTION ADDED 19 SEPTEMBER 2026 ------------------------------------------------------------------------------ NHTSA stores complaint narratives in ALL CAPS for records filed through roughly 2020 and in mixed case for records filed from 2021 on. A case-SENSITIVE search of the narrative field loses about 92% of the complaints filed after 2020, so this is the single largest way an analysis of this database can go wrong while still looking finished. Every query behind this file is case-insensitive. SQLite's LIKE and its FTS5 full-text index both ignore case by default, and no query in this work uses GLOB or any case-sensitive comparison. This was not taken on trust. On 19 September 2026 every published folder was re-checked against the same database, comparing the figures here with what a case-sensitive query would have returned. The result: the complaint set published here (5,009) is LARGER than the absolute ceiling a case-sensitive search of the same terms could have returned (1,953 narratives). A case-sensitive query cannot produce a bigger set than itself, so the bug is ruled out for this folder by arithmetic, not by assurance.