PulsCar analysis of NHTSA public data, September 2026 Stalling while driving: 36,754 owner complaints, 556 recall campaigns, 11,625 manufacturer bulletins Licensed CC BY 4.0. Query run 18 September 2026; the 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 (23,187) 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 and the page described a 2015-2020 file while claiming the full window. Re-running the same rules case-insensitively returns 36,754 complaints, spread across every year, and every complaint figure on this page, in the article and in the CSVs has been recomputed from that set. What changed materially: the tow share rose from 8.21% to 11.42%, the crash share fell from 2.15% to 1.87%, the fire share from 2.03% to 1.78%; the speed distribution shifted slightly upward (40.2% at 46 mph or above against 38.0%); the repair-verb subset grew from 5,473 to 9,893; Hyundai and Kia move up the by-make table (Hyundai 2,939 and Kia 2,688 against 1,322 and 1,281), which is what the missing post-2020 years were hiding. The recall (556) and bulletin (11,625) counts are NOT affected: those queries matched NHTSA's recall and bulletin text, which is mixed case throughout the window, and re-running them case-insensitively returns the same order of magnitude (a case-sensitive match on the bulletin file returns 1,554, which is not what was published). One caveat about this correction, stated because it is real: the rebuild re-implements the rules documented below from this file. Run case-sensitively it returns 21,729 complaints against the 23,187 published earlier, so it is a faithful but not byte-identical reconstruction of the original script (6% tighter); the difference is in regex detail the original script, not this file, held. The subset and repair-category regexes were likewise re-derived from the descriptions below, so those figures are new measurements of the same intent rather than the identical measurement. -------------------------------------------------------------------------- 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 -- campaign, make, model, model year, component, manufacturer, potentially affected, report date, defect, consequence, remedy. NHTSA manufacturer communication files TSBS_RECEIVED_2015-2019, _2020-2024, _2025-2026 -- https://static.nhtsa.gov/odi/ffdd/tsbs/ (dictionary TSBS.txt) -------------------------------------------------------------------------- WHICH COMPLAINTS COUNT (the stall-while-driving set: 36,754) -------------------------------------------------------------------------- A complaint counts when its narrative contains ONE SENTENCE in which a stall word and a driving word appear within 90 characters of each other. ALL MATCHING IS CASE-INSENSITIVE (Python re.IGNORECASE); the word lists are printed here in capitals only for readability. stall words: STALL | STALLS | STALLED | STALLING | SHUT OFF | SHUT DOWN | SHUTS OFF | SHUTS DOWN | SHUT ITSELF OFF | CUT OFF | CUT OUT | CUTS OFF | CUTS OUT | DIED | DIES | DIE | TURNED OFF | TURNS OFF | LOST POWER AND STOPPED | ENGINE (JUST) QUIT | QUIT RUNNING driving words: WHILE DRIVING | WAS DRIVING | DRIVING DOWN/ON/AT/ALONG/ HOME/TO | ON THE HIGHWAY/FREEWAY/INTERSTATE/ROAD/HWY | AT MPH | IN TRAFFIC | MERGING | IN MOTION | MID-TURN | GOING | ACCELERAT* | IN THE MIDDLE OF THE ROAD/HIGHWAY/FREEWAY/INTERSECTION | CROSSING | MAKING A LEFT/RIGHT TURN | ON THE ON-RAMP | ON THE OFF-RAMP EXCLUSIONS, applied per sentence: no-start and parked narratives -- WOULD NOT START, WON'T START, DID NOT START, FAILED TO START, NO-START CONDITION, WHILE PARKED, IN THE DRIVEWAY, AFTER I PARKED; auto stop-start and remote start -- AUTO STOP, AUTO START-STOP, START-STOP, REMOTE START (the engine switching itself off at a light by design is not a stall); denials -- DOES NOT STALL, NEVER STALL, DID NOT STALL. Makes excluded: motorcycle, heavy-truck, bus and RV manufacturers (same list as our other analyses: Harley-Davidson, Indian, Triumph, Ducati, Kawasaki, Yamaha, Suzuki, Polaris, Can-Am, KTM, Vespa, Piaggio, Aprilia, Moto Guzzi, Arctic Cat, Royal Enfield, Husqvarna, Zero, BMW Motorrad, Mack, Hino, Kenworth, Peterbilt, Paccar, International, IC Bus, Freightliner, Western Star, Blue Bird, Thomas Built, Navistar, Prevost, Workhorse, Spartan, Autocar, Forest River, Volvo Trucks, Volvo Bus, Isuzu Trucks, UD Trucks, Sterling, Gillig). SAMPLE READING (required before publishing any keyword count) Thirty randomly drawn matches were read in full. Twenty-eight described an engine that stopped running while the vehicle was moving. Two did not: one reported the instrument cluster and infotainment shutting off while driving with no statement that the engine stopped, and one reported a three-second electrical reset. FALSE-POSITIVE RATE: 2 IN 30, ROUGHLY 7%. The dominant residual error is electrical-blackout narratives being read as engine stalls; the counts below should be read with that in mind. -------------------------------------------------------------------------- FIELDS AND RESULTS -------------------------------------------------------------------------- Outcome flags (share answering Y, of 36,754): towed 11.42%, crash 1.87%, fire 1.78%, 517 injuries, 17 deaths. All-complaint baseline (777,040): towed 5.11%, crash 4.37%, fire 1.82%. Mileage: NHTSA MILES field, rows with 0 < miles < 400,000 (24,764 rows). Median 79,000; quartiles 43,000 and 115,000; 21.4% at or under 36,000 miles; 36.6% at or under 60,000. SPEED: NHTSA SPEED field, rows with 0 < speed < 120 mph (26,227 rows). Median 40 mph; quartiles 25 and 60. 1-15 mph 3,814 (14.5%); 16-30 mph 5,112 (19.5%); 31-45 mph 6,748 (25.7%); 46-60 mph 5,171 (19.7%); 61 mph and above 5,382 (20.5%). LIMITATION: the speed field is the owner's own entry, it is blank in about 21% of the set, and NHTSA's contact-report format prompts for it, so complaints filed by phone are over-represented among those that carry one. Phrase subsets, share of the full 36,754 narratives (regexes re-derived 18 September 2026, see CORRECTION): "without warning" / "no warning" / "suddenly" 10,696 (29.1%); restarted or would restart 10,619 (28.9%); highway / freeway / interstate 9,856 (26.8%); mentions a recall 9,943 (27.1%); check engine light or MIL 5,126 (13.9%); loss of power steering 631 (1.7%) -- all-complaint baseline 1.03%; loss of braking assist 253 (0.7%) -- all-complaint baseline 0.67%; "no fault found" family 1,239 (3.4%) -- all-complaint baseline 1.56%. The three all-complaint baselines were themselves recomputed case-insensitively on 18 September 2026; the earlier 0.51% and 0.95% were uppercase-only matches over the same 777,040 rows and understated the file. The "no fault found" family is: could not duplicate / replicate / reproduce / find, unable to duplicate / replicate / reproduce, no problem / fault / issue / trouble found, operating as designed, within specification, nothing found, no codes stored / found / present. WHAT OWNERS SAID WAS REPAIRED (9,893 complaints) A complaint enters this table when the narrative contains a repair verb (DIAGNOS*, DETERMINED, REPLACE*, REPAIR*, FIXED, INSTALL*, NEEDED A NEW, TOLD ME/US THE) AND names at least one part family. Categories overlap, so shares do not sum to 100. Full regexes are visible in the category names in stall-what-owners-said-was-repaired.csv; the two that need explaining: - "Battery replaced" matches the word BATTERY anywhere in a narrative that also names a repair. It therefore captures both the owners whose stall really was a battery and the much larger group who had a battery fitted as the first guess. It is published as a measure of what gets replaced first, not of what causes stalls. - "Alternator, charging system or voltage regulator" is a separate, narrower category (851 complaints) and does not include the word battery. This is NOT a cause ranking. It counts which stalls end in a federal safety complaint, which is biased towards expensive and repeated failures. RECALL CAMPAIGNS (556) Every campaign whose defect or consequence text names a stall, an engine shutdown while driving, a loss of motive power or a loss of propulsion, after excluding the motorcycle, bus and heavy-truck makers above. 556 distinct campaigns. Summing each campaign's potentially-affected count ONCE (the flat file repeats it on every make/model/year row) gives 37,441,163 vehicles. That total double counts vehicles covered by more than one campaign -- the 2011-2013 Jeep Grand Cherokee and Dodge Durango, for instance, appear in 14V-530 or 15V-115 and again in 19V-813 -- so read it as "campaign-vehicles", not as distinct cars. 119 of the 556 campaigns (21.4%) have a remedy whose text mentions software, reprogramming, a reflash, a calibration or an update. Campaign-level rows are published in stall-recall-campaigns.csv. MANUFACTURER BULLETINS (11,625) Distinct NHTSA document IDs whose Summary names a stall, stalling, an engine shutdown while driving, a loss of motive power or propulsion, or an engine that cuts out. One bulletin = one document ID; the flat file's one row per make/model/year is collapsed. Same maker exclusions, plus Cummins and FCCC. 1,984 (17.1%) mention software, reprogramming, a reflash or a calibration. Per-group counts: GM 2,080 (16% software), Toyota 1,242 (8%), Stellantis 877 (22%), Ford 752 (24%), VW Group 617 (26%), BMW 517 (5%), Kia 457 (45%), Mercedes-Benz 413 (31%), Nissan 372 (15%), Porsche 366 (5%), Mazda 341 (8%), Honda 294 (11%), Subaru 282 (53%), Volvo 239 (22%). A bulletin covering several groups is counted in each group's figure. -------------------------------------------------------------------------- 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, so nobody knows how many of each model are on the road. 2. Counts favour high-volume models and models whose problem has been publicized. Nothing here ranks reliability, and no honest reading of complaint counts can. 3. The keyword method carries roughly one mis-sort in fourteen (see SAMPLE READING). Counts under about 30 should be read as "a handful of owners". 4. Towed, crash, fire, mileage and speed are the owner's own entries. The towed and crash shares of any complaint set are inflated relative to the real world, because a frightening event is what prompts a filing. The crash share here (1.87%) is LOWER than the all-complaint baseline, which is a property of the baseline -- crash-heavy components such as airbags and seat belts pull it up -- not evidence that stalls are safe. 5. Recall "potentially affected" is the manufacturer's estimate of vehicles in the population, not of vehicles that have the defect. Where a Part 573 report gives an estimated percentage with the defect, the article says so. 6. "Mentions software" counts the word, not the outcome. -------------------------------------------------------------------------- FILES -------------------------------------------------------------------------- stall-complaints-by-make.csv make, complaints, miles reported, median miles, share at or under 36,000 miles, percent towed, percent crash, speeds reported, median speed. 10-complaint floor. stall-top-models.csv same columns plus top model year and its count, for the 80 most-reported models. stall-speed-distribution.csv speed band, complaints, share of the 26,227 complaints that report a speed. stall-what-owners-said-was-repaired.csv repair category, complaints naming it, share of the 9,893 that name a repair. stall-recall-campaigns.csv campaign, manufacturer, report date, potentially affected, makes, whether the remedy text mentions software. 556 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.