PULSCAR ANALYSIS — P0300 IN NHTSA OWNER COMPLAINTS: WHO REPORTS A RANDOM MISFIRE, WHERE THEY STAND ON THE WARRANTY CLOCK, WHAT THE REPORT SAYS WAS FOUND, AND WHAT IT COST PulsCar analysis of NHTSA public data, September 2026. License: CC BY 4.0. Attribution: "PulsCar analysis of NHTSA public data, September 2026." Published alongside https://www.pulscar.io/blog/p0300-code ------------------------------------------------------------------------------ 0. WHAT THIS FILE ADDS, AND WHAT IT DELIBERATELY DOES NOT RECOMPUTE ------------------------------------------------------------------------------ The PulsCar Check Engine Code Index already counts the manufacturer bulletins that name P0300 — 431 distinct bulletins from 14 manufacturer groups, at least 44% of them software updates (20% with Stellantis set aside) — and publishes its own method and data: https://www.pulscar.io/blog/check-engine-code-index Those bulletin figures are reused unchanged here and in the article. Nothing in this file recounts them. This file answers the complaint side, which the index does not: who files a P0300 report, how old the car is and how far it has run, what the report says a shop eventually found, how often a shop looked and found nothing, what owners were quoted or paid, and how rarely a federal safety recall is involved at all. ONE ARITHMETIC DIFFERENCE, STATED OPENLY. The index reports 121 odometer readings in the P0300 complaint set and a median of 77,398 miles. Our own extraction keeps every reading above zero and below 400,000 and returns 122, including one report whose odometer field reads 6 miles, which gives a median of 77,199. Dropping that one implausible entry leaves 121 readings and reproduces the index figure exactly, so the difference is a filtering convention on data-entry artifacts and not a disagreement about the data. The article quotes the index figure, 77,398, so that the two pages agree. ------------------------------------------------------------------------------ 1. SOURCE FILES AND DATE RANGE ------------------------------------------------------------------------------ COMPLAINTS NHTSA ODI complaint flat files COMPLAINTS_RECEIVED_2015-2019, _2020-2024 and _2025-2026. Field layout: https://static.nhtsa.gov/odi/ffdd/cmpl/CMPL.txt 777,040 distinct vehicle complaints (product type V), de-duplicated by ODI report number, received 1 January 2015 to 8 September 2026. RECALLS NHTSA recall flat file FLAT_RCL, campaigns with report dates from 1 January 2010 to 9 September 2026; 15,241 distinct campaigns. INVESTIGATIONS NHTSA ODI investigation flat file FLAT_INV. Field layout: https://static.nhtsa.gov/odi/ffdd/inv/INV.txt Download: https://static.nhtsa.gov/odi/ffdd/inv/FLAT_INV.zip Data pulled 14 September 2026; recall and investigation tables current to 9 September 2026. This analysis run 18 September 2026. No third-party data. ------------------------------------------------------------------------------ 2. THE COMPLAINT SET ------------------------------------------------------------------------------ Query: the owner narrative (DESCR) matched against the whole-word regular expression \bP0300\b, case-insensitive. SELECT * FROM complaints WHERE descr LIKE '%P0300%'; -- 385 rows then kept only rows matching \bP0300\b -- 365 rows FINAL SET: 365 complaints. This is the denominator for every percentage below unless another denominator is named. The same whole-word rule is used by the Check Engine Code Index, so the two sets are identical. THE 20 REJECTED ROWS WERE READ, ONE BY ONE. This is where the whole-word rule costs something, and the honest result is that it costs more than it does for other codes. Nineteen of the twenty are genuine P0300 reports whose code is run together with something else: manufacturer sub-codes ("P030085", "P030027", "P030002" — the two-digit failure-type suffix Toyota, Lexus and Mercedes-Benz scan tools append), doubled zeros ("P030000", "P03000"), or a range written with an en dash ("P0300-P0306", "P0300-P0304"). One is a mistyped cylinder code ("P03003 (cylinder 3)"). So 365 is a floor: a looser rule would return about 384. We keep 365 because it is the number the index publishes and because a looser rule also drags in P0301-P0308 ranges; the undercount is named in the article rather than silently absorbed. A separate set, P0300 through P0308 matched the same way, contains 1,394 complaints and is used only for the Ford coolant-intrusion comparison in section 6. Self-reported, no denominator, not a failure rate. A complaint set is drawn from owners angry or worried enough to file with the federal government. It says nothing about how common a fault is per vehicle sold, and nothing here should be read as a reliability ranking. Sales volumes differ by an order of magnitude between the nameplates in this file. ------------------------------------------------------------------------------ 3. WHAT THE 365 LOOK LIKE (p0300-complaints-by-make-and-model.csv, p0300-warranty-position.csv) ------------------------------------------------------------------------------ BY MAKE (top): Ford 82, Chevrolet 66, Honda 42, GMC 27, Kia 26, Hyundai 20, Jeep 17, Nissan 9, Ram 9, Buick 7, Dodge 7, Chrysler 7, Toyota 7, Cadillac 6. BY MODEL (top): Ford Escape 35, Ford Fusion 19, Chevrolet Silverado 1500 18, Honda Accord 14, Kia Soul 13, GMC Sierra 1500 11, Chevrolet Tahoe 10, Honda CR-V 10, Ford Edge 9, Chevrolet Traverse 8, Ram 1500 8. SEVERITY FLAGS, against the whole 777,040-complaint file as the baseline: P0300 set all complaints crash flagged 0 (0.00%) 33,940 (4.37%) fire flagged 0 (0.00%) 14,121 (1.82%) towed 10 (2.74%) 39,739 (5.11%) any injury 0 21,897 any death 0 Zero is a real zero here, not a missing field: the same three flags are present and populated across the file. Read it as what it is — an emissions-and-wallet code that owners nonetheless file as a safety complaint because the car loses power. It is not evidence that a misfire is harmless; a flashing light on a highway is described in dozens of these narratives. VEHICLE AGE AT REPORT (report year minus model year; 365 rows, all usable): 0-3 years 67 · 4-7 years 179 · 8-11 years 96 · 12+ years 23. Median 6 years. ODOMETER AT REPORT (122 of 365 state one; readings at or below 0 and at or above 400,000 excluded — none were): under 36,000 31 · 36,000-59,999 14 · 60,000-99,999 39 · 100,000+ 38. Median 77,398 (index convention; see section 0). Middle half roughly 34,300 to 110,540. Lowest 6 miles, highest 325,599. REPORT YEAR: 2015 9 · 2016 14 · 2017 11 · 2018 20 · 2019 16 · 2020 22 · 2021 16 · 2022 24 · 2023 35 · 2024 52 · 2025 68 · 2026 78. 2026 covers 1 January to 8 September only, so it is a partial year that has already passed every full year before it. ------------------------------------------------------------------------------ 4. READING 120 COMPLAINTS END TO END (p0300-120-complaints-read.csv) ------------------------------------------------------------------------------ Keyword tallies over the 365 are reported in the article where they are defensible, but the question "what was it, in the end" cannot be answered by keyword. So a random sample was drawn and read in full. SAMPLE: Python random.sample, seed 20260918, 120 of the 365, sorted by report number. Every narrative was read to its end (NHTSA truncates long narratives; the first 1,400 characters were used, which covers the outcome in all but a handful, and those are coded "no diagnosis reached"). EACH REPORT WAS GIVEN ONE PRIMARY OUTCOME — what the narrative says a shop found or did, not what the owner guesses. Where several things happened, the outcome is the last confirmed finding (a car that got plugs, then a borescope, then a coolant finding is coded coolant, and separately flagged as an ignition-first purchase). coolant intrusion, head gasket or cracked head 24 (20.0%) internal mechanical: pistons, rings, bearings, low compression or a seized engine 24 (20.0%) valvetrain: lifter, camshaft, rocker, pushrod, timing 13 (10.8%) fuel delivery: pump, injector or fuel line 10 (8.3%) ignition parts fitted (plugs, coils, boots) 10 (8.3%) intake, PCV, EVAP, throttle body or intake-valve carbon 6 (5.0%) software update, recall or extension work performed 4 (3.3%) no diagnosis reached, could not duplicate, not repaired 29 (24.2%) --- 120 THREE FLAGS, each read from the text: A shop looked at the car and could not find or reproduce the fault 11 (9.2%) Plugs and/or coils were fitted and the misfire or code came back 13 (10.8%) A named recall, bulletin, campaign or extension did NOT cover this car 19 (15.8%) A manufacturer program or warranty paid for some or all of the work 6 (5.0%) FALSE-POSITIVE AND FALSE-NEGATIVE CHECK ON THE "NO FAULT FOUND" KEYWORD QUERY. The regular expression could not (be )?duplicat|cannot duplicat|no (problem|fault|issue|codes?) (was )? found|found nothing|unable to (duplicate|replicate|find)|nothing wrong| could not find (the |any )?(problem|issue|cause) returns 9 of the 365 (2.5%), against 1.51% across all 777,040 complaints. All 9 were read: 9 of 9 are genuine (no false positives). But hand-reading 120 found 11 (9.2%) that describe a shop looking and finding nothing, in wordings the expression cannot catch: "they now cannot find out what the problem is", "they could not/would not", "both dealerships said there were no codes and they didn't know what was wrong", "they cannot get the Engine Light to illuminate", "was told everything looked fine". The keyword figure therefore undercounts by roughly a factor of three, and the article publishes the hand-read figure with its sample size, not the keyword figure. LIMITS OF THE CODING. One reader, one pass, no second coder, so borderline cases (is a "burnt valve" valvetrain or internal mechanical?) were resolved by a written rule: anything about how a valve is opened is valvetrain, anything about sealing or the parts below the head is internal mechanical. A 120-row sample gives a 95% confidence interval of roughly ±7 percentage points on a 20% share; treat every share here as a rough proportion, not a measurement. ------------------------------------------------------------------------------ 5. WHAT OWNERS SAY THEY WERE QUOTED OR PAID (p0300-owner-reported-repair-figures.csv) ------------------------------------------------------------------------------ All 365 narratives were scanned for dollar amounts with the expression \$\s?([\d][\d,]{2,9})(\.\d\d)?, keeping values from $50 to $60,000: 69 complaints carry at least one figure, 97 figures in all. Every figure was then read in its own sentence, and only figures for an engine, short block, long block, head gasket, cylinder head, or a camshaft-and-lifter job were kept. Excluded by reading: vehicle values and trade-in offers ("a $6,500 quote on an $8,000 vehicle" keeps 6,500 and drops 8,000), diagnostic and teardown fees, extended-warranty purchase prices, oil changes, and one cumulative multi-year total. Where one complaint carries two options for the same car, the engine figure is kept and the alternative is written into the row description, so each row is one household. RESULT: 29 figures. Median $9,000. Range $2,000 to $18,000. Six of the 29 are sums actually paid; the other 23 are quotes or estimates the owner declined, could not afford, or had not yet accepted. This is not a price survey. It is what 29 angry owners told the federal government about the top of the P0300 ladder, and it is published only to show the shape of that top end. The article marks it as an owner-reported range. ------------------------------------------------------------------------------ 6. THE FORD COOLANT CLUSTER, AND THE GM VALVETRAIN CLUSTER ------------------------------------------------------------------------------ Of the 365, 78 mention coolant or antifreeze. 64 of those 78 are Ford and 3 more are Lincoln: 67 of 78, or 86%, are one manufacturer group. By model: Escape 29, Fusion 19, Edge 9, Mustang 5, Lincoln MKC 2, MKZ 1. In the wider P0300-P0308 set of 1,394, the Ford Escape, Edge and Fusion together account for 452 reports, and 395 of those 452 mention coolant or antifreeze. These match Ford's own coolant-intrusion bulletins by model and year, and they are the reason the article treats coolant as a first-rank question rather than a footnote. Of the 365, 38 mention a lifter, camshaft, cam shaft, pushrod or push rod. 26 of the 38 (68%) are GM marques: Chevrolet 12, GMC 12, Buick 1, Cadillac 1. By model the top entries are Silverado 1500 6, Sierra 1500 5, Yukon XL 5, Tahoe 2. Again: complaint counts, not failure rates, and both clusters are also the highest-volume vehicles in their segments. ------------------------------------------------------------------------------ 7. RECALLS AND INVESTIGATIONS THAT NAME A MISFIRE ------------------------------------------------------------------------------ Query: the recall Component, Defect, Consequence, Remedy and Notes fields matched against /misfir/i, counted as distinct campaign numbers. RESULT: 27 campaigns in the whole file, from 1 January 2010 to 9 September 2026. Twenty-four of the 27 are heavy trucks, transit and school buses, motorcycles or medium-duty chassis — Cummins Westport, PACCAR, Mack, Daimler Trucks, New Flyer, Blue Bird, NABI, Gillig, Autocar, Oshkosh, Navistar, Eldorado, Crane Carrier, Motor Coach Industries, Kalmar, Isuzu commercial trucks, Polaris Victory and Indian motorcycles, and one Starcraft transit bus. THREE are light passenger vehicles: 12V-107 Porsche, 1,232 units, 2012 911 Carrera S. A coolant line rubbing a fuel line; a disconnected fuel line "could lead to the engine misfiring or stalling". Remedy: replace the fuel line. 18V-636 Chrysler (FCA US), 34,357 units, 2017-2018 Alfa Romeo Giulia and 2018 Stelvio with the 2.0L. "These engines may misfire causing the catalytic converter to overheat." Remedy: inspect for codes and update the engine control module software. 26V-493 Hyundai, 42 units, 2027 Genesis G80. "The ignition coil ground bolt may be loose, resulting in an engine misfire or stall while driving." Remedy: tighten the bolt. Owner letters expected 27 September 2026. Campaign text verified live against https://api.nhtsa.gov/recalls/campaignNumber?campaignNumber=18V636000 and the same endpoint for 26V493000 and 12V107000. That is the finding the article leads with in its recall section: across sixteen and a half years and 15,241 campaigns, the misfire that fills 431 manufacturer bulletins produced three light-vehicle safety recalls totalling about 35,600 cars. The repairs behind P0300 live in bulletins and warranty extensions, which carry clocks, rather than in recalls, which do not. INVESTIGATIONS naming a misfire: 6 distinct ODI actions. DP09-004 Ford, engine stalling, 2009-2010 (closed) PE09-032 General Motors, engine compartment fire, 2009 (closed) PE13-018 Ford, reduced power during hard acceleration, 2011-2013 F-150 3.5L EcoBoost, opened 22 May 2013, closed 7 April 2014 PE17-001 Suzuki, motorcycle stalling (closed) PE21-008 Volkswagen, fuel injector leak, 2018-2020 Atlas and Atlas Cross Sport 3.6L, opened 29 March 2021, closed 3 October 2022 DP26-008 Ford, "Loss of Motive Power", 2015-2018 Edge 2.0L EcoBoost, defect petition received 5 August 2026, OPENED 3 SEPTEMBER 2026 and still open at the date of this file. ODI's own words: the petition "alleges that engine coolant intrusion results in a significant loss of engine power" and that "after the vehicle has sat and cooled down, the engine runs roughly and generates a check engine light for a cylinder misfire." The petition is filed as ODI number 11754989. A defect petition being opened decides nothing; NHTSA will grant or deny it. ------------------------------------------------------------------------------ 8. WHAT THE FILES CONTAIN ------------------------------------------------------------------------------ p0300-complaints-by-make-and-model.csv make, model, complaints, complaints_for_this_make, share_of_365_pct p0300-warranty-position.csv measure (vehicle_age_at_report | odometer_at_report | report_year), band, complaints, denominator p0300-120-complaints-read.csv n, odi_report_number, model_year, make, model, odometer_reported, what_the_report_says_was_found_or_done, shop_looked_and_found_nothing, ignition_parts_fitted_and_fault_returned, named_document_did_not_cover_this_car, manufacturer_program_paid p0300-owner-reported-repair-figures.csv odi_report_number, model_year, make, model, amount_usd, what_the_amount_is_for, paid_or_quoted No personal data is published: no VIN, no name, no city, no dealer name, and no free text beyond the short repair description written by us. Every row can be traced by its ODI report number at nhtsa.gov/complaints. ------------------------------------------------------------------------------ 9. LIMITATIONS, IN ONE PLACE ------------------------------------------------------------------------------ * Complaints are self-reported and unaudited. They are not failure rates and carry no denominator of vehicles on the road. * 365 is a small set. Every model-level count in it is small enough that a handful of reports moves the ranking. * Only 122 of 365 state an odometer reading, and owners round. * The 120-row hand coding is one reader's judgment, single pass. * Dollar figures are what owners wrote, not invoices we have seen. * Recall and bulletin coverage always depends on the VIN and on a date, and both change. Every clock in the article was computed from the document's own terms against September 2026 and will be wrong later.